Matthew Jordan [Mon, 30 Jun 2014 03:25:59 +0000 (03:25 +0000)]
chan_sip: be more tolerant of whitespace between attributes in SDP fmtp line
This patch is essentially a backport of a small portion of r397526 from
ASTERISK-21981. In that patch, pass through support and format attribute
negotiation was added for Opus. Part of that included being more tolerant to
whitespace in the fmtp line of an SDP; that part of the patch is being
applied here.
As the author of the backport pointed out, in SDP, the fmtp line is allowed to
include whitespace between attributes. RFC 3267 chapter 8.3 (from 2001)
includes an example for this. This was not removed in the updated RFC 4867 in
2007.
Review: https://reviewboard.asterisk.org/r/3658
#ASTERISK-23916 #close
Reported by: Alexander Traud
patches:
sdpFMTPspace_Asterisk11.patch uploaded by Alexander Traud (License 6520)
........
Merged revisions 417587 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........
Merged revisions 417588 from http://svn.asterisk.org/svn/asterisk/branches/11
Richard Mudgett [Fri, 27 Jun 2014 23:11:26 +0000 (23:11 +0000)]
event.c: Fix type mismatch errors in ie_maps[].
In v12+ the type values from the table are only used by the CEL unit
tests. Since the unit tests were only comparing a generated expected
event with a real event to see if the ie contents matched and using the
same table IE_PLTYPE values to read the event contents, the type
mismatches were not detected.
Corey Farrell [Fri, 27 Jun 2014 19:27:13 +0000 (19:27 +0000)]
Ensure REF_DEBUG records entrys for attempts to ao2_ref an invalid object
This change ensures that __ao2_ref_debug writes to ref_log when given a
non-NULL pointer to an invalid ao2 object. This is to ensure that we
record any attempt manipulate references of already freed objects.
Matthew Jordan [Fri, 27 Jun 2014 13:48:12 +0000 (13:48 +0000)]
res_pjsip: Add ActionID to events created as a result of PJSIP AMI actions
A number of various PJSIP AMI actions were failing to parse out and place the
ActionID into their responses. This patch updates the various PJSIP actions
such that the passed in ActionID is emitted on any event list complete events,
as well as any intermediate events created as a result of the action.
Matthew Jordan [Thu, 26 Jun 2014 12:23:27 +0000 (12:23 +0000)]
udptl: Correct FEC to not consider negative sequence numbers as missing
When using FEC, with span=3 and entries=4 Asterisk will attempt to repair
the packet with sequence number 5, as it will see that packet -4 is
missing. The result is Asterisk sending garbage packets that can kill a
fax.
This patch adds a check to see if the sequence number is valid before
checking if the packet is missing.
Matthew Jordan [Thu, 26 Jun 2014 12:10:11 +0000 (12:10 +0000)]
res_http_websocket: Close websocket correctly and use careful fwrite
When a client takes a long time to process information received from Asterisk,
a write operation using fwrite may fail to write all information. This causes
the underlying file stream to be in an unknown state, such that the socket
must be disconnected. Unfortunately, there are two problems with this in
Asterisk's existing websocket code:
1. Periodically, during the read loop, Asterisk must write to the connected
websocket to respond to pings. As such, Asterisk maintains a reference to
the session during the loop. When ast_http_websocket_write fails, it may
cause the session to decrement its ref count, but this in and of itself
does not break the read loop. The read loop's write, on the other hand,
does not break the loop if it fails. This causes the socket to get in a
'stuck' state, preventing the client from reconnecting to the server.
2. More importantly, however, is that the fwrite in ast_http_websocket_write
fails with a large volume of data when the client takes awhile to process
the information. When it does fail, it fails writing only a portion of
the bytes. With some debugging, it was shown that this was failing in a
similar fashion to ASTERISK-12767. Switching this over to ast_careful_fwrite
with a long enough timeout solved the problem.
Note that this version of the patch, unlike r417310 in Asterisk 11, exposes
configuration options beyond just chan_sip's sip.conf. Configuration options
to configure the write timeout have also been added to pjsip.conf and ari.conf.
Corey Farrell [Wed, 25 Jun 2014 18:56:09 +0000 (18:56 +0000)]
ao2_container node object ignores REF_DEBUG in all places except one
Almost every reference operation against container node's uses
__ao2_alloc or __ao2_ref, thereby preventing ref logging for
the nodes. One node reference is released with ao2_t_ref, causing
refcounter.py to falsely report skews and leaks for many nodes.
Richard Mudgett [Mon, 23 Jun 2014 15:53:09 +0000 (15:53 +0000)]
core_unreal: Fix off by one buffer overwrite error.
Appending the ;2 to the user supplied ;1 uniqueid to create the ;2 version
if the user did not also supply an extra uniqueid for the ;2 channel
resulted in allocating a buffer that was one byte too small.
* Fix off by one error in ast_unreal_new_channels() when generating the ;2
uniqueid from the user suppled ;1 version.
* Pulled some long assignment lines from if tests to improve line break
readability in ast_unreal_new_channels().
George Joseph [Fri, 20 Jun 2014 23:16:43 +0000 (23:16 +0000)]
build: Allow autoconf/ast_ext_tool_check to handle cross-compiling better.
ast_ext_tool_check.m4 isn't handling cases where a path to a package is
provided (E.G. --with-mysqlclient=/some/sysroot) and the package has a config
tool (E.G. mysql_config) and the package has its own subdirectories in include
or lib. For example, mysql's libraries are in ${MYSQLCLIENT_DIR}/usr/lib/mysql
but ast_ext_tool_check sets MYSQLCLIENT_LIB to ${MYSQLCLIENT_DIR}/usr/lib.
libxml2 has the same problem with its includes. They're in
${LIBXML2_DIR}/usr/include/libxml2 not directly in ${LIBXML2_DIR}/usr/include.
Both cause configure to fail and there are others in the same boat.
The problem is caused by logic in ast_ext_tool_check that overrides the result
of the config tool's --cflags and --libs options if package_DIR is set.
This patch prepends package_DIR (if specified) to the -L and -I results from
the package's config tool instead of overriding them.
A regenerated ./configure and include/asterisk/autoconfig.h.in are included
but can be regenerated by running ./bootstrap.sh at any time.
Tested by: George Joseph
Tested by: Matt Jordan
Review: https://reviewboard.asterisk.org/r/3550/
........
Merged revisions 416929 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........
Merged revisions 416930 from http://svn.asterisk.org/svn/asterisk/branches/11
George Joseph [Fri, 20 Jun 2014 21:57:57 +0000 (21:57 +0000)]
build: Allow autoconf/ast_ext_tool_check to handle cross-compiling better.
ast_ext_tool_check.m4 isn't handling cases where a path to a package is
provided (E.G. --with-mysqlclient=/some/sysroot) and the package has a config
tool (E.G. mysql_config) and the package has its own subdirectories in include
or lib. For example, mysql's libraries are in ${MYSQLCLIENT_DIR}/usr/lib/mysql
but ast_ext_tool_check sets MYSQLCLIENT_LIB to ${MYSQLCLIENT_DIR}/usr/lib.
libxml2 has the same problem with its includes. They're in
${LIBXML2_DIR}/usr/include/libxml2 not directly in ${LIBXML2_DIR}/usr/include.
Both cause configure to fail and there are others in the same boat.
The problem is caused by logic in ast_ext_tool_check that overrides the result
of the config tool's --cflags and --libs options if package_DIR is set.
This patch prepends package_DIR (if specified) to the -L and -I results from
the package's config tool instead of overriding them.
Tested by: George Joseph
Tested by: Matt Jordan
Review: https://reviewboard.asterisk.org/r/3550/
........
Merged revisions 416870 from http://svn.asterisk.org/svn/asterisk/branches/11
George Joseph [Fri, 20 Jun 2014 15:22:44 +0000 (15:22 +0000)]
astobj2: Additional refactoring to push impl specific code down into the impls.
Move some implementation specific code from astobj2_container.c into
astobj2_hash.c and astobj2_rbtree.c. This completely removes the need for
astobj2_container to switch on RTTI and it no longer has any knowledge of
the implementation details.
Also adds AO2_DEBUG as a new compile option in menuselect which controls
astobj2 debugging independently of AST_DEVMODE and REF_DEBUG.
Tested by: George Joseph
Review: https://reviewboard.asterisk.org/r/3593/
George Joseph [Thu, 19 Jun 2014 20:12:15 +0000 (20:12 +0000)]
pjsip cli: Change Identify to show CIDR notation instead of netmasks.
* Added ast_sockaddr_cidr_bits() to count the 1 bits in an ast_sockaddr.
* Added ast_ha_join_cidr() which uses ast_sockaddr_cidr_bits() for the netmask
instead of ast_sockaddr_stringify_addr.
* Changed res_pjsip_endpoint_identifier_ip to call ast_ha_join_cidr() instead
of ast_ha_join() for the CLI output.
This is a CLI change only. AMI was not affected.
Tested by: George Joseph
Review: https://reviewboard.asterisk.org/r/3652/
George Joseph [Thu, 19 Jun 2014 16:03:17 +0000 (16:03 +0000)]
Remove the problematic and unneeded AST_MODFLAG_GLOBAL_SYMBOLS from pbx_lua.c
AST_MODFLAG_GLOBAL_SYMBOLS was causing the module to be incorrectly loaded
before pbx_config. pbx_config was therefore blowing away contexts that were
created by pbx_lua. With AST_MODFLAG_DEFAULT the load order is now correct
and contexs are being properly merged. AST_MODFLAG_GLOBAL_SYMBOLS was not
needed anyway since no other modules needed its global symbols that early.
ASTERISK-23818 #close
Reported by: Dennis Guse
Tested by: Dennis Guse
Tested by: George Joseph
George Joseph [Wed, 18 Jun 2014 17:09:04 +0000 (17:09 +0000)]
Update extensions.lua.sample with naming conflict guidance.
The sample extensions.lua was causing pbx_lua to fail to load when parsing
'app.goto("default", "s", 1)' because in Lua 5.2, 'goto' is now a reserved
word. This patch adds guidance to extensions.lua.sample and changed
'app.goto("default", "s", 1)' to 'app.['goto']("default", "s", 1)'.
Matthew Jordan [Wed, 18 Jun 2014 04:16:55 +0000 (04:16 +0000)]
stasis_channels: Update the stasis cache if manager variables are needed
In r416211, the publishing of variable changes was modified such that a
cached channel snapshot was used if manager variables were not requested
with each AMI event. This was done to reduce the amount of channel snapshots
created.
However, an assumption was made that generating a channel snapshot and
publishing the snapshot to the channel topic was sufficient to ensure that
the cache would be updated; this is not the case. The channel snapshot type
must be used to force a snapshot update.
This patch updates the publication of channel variables such that the cache
is updated prior to publication of the channel variable message if manager
variables are in use. This ensures that all AMI events receive the variable
update when they are supposed to.
Note that this issue was caught by the Asterisk Test Suite (go go testing)
Kinsey Moore [Tue, 17 Jun 2014 16:26:10 +0000 (16:26 +0000)]
MoH: Don't restart stream on repeated start calls
Currently, music on hold will stop and then start again from the
beginning if ast_moh_start() is called multiple times. This can happen
if a call is put on hold repeatedly (the channel receives multiple
HOLD control frames) and can be triggered from ARI by starting MoH on a
channel multiple times. This is fairly jarring/annoying to users.
This change prevents MoH from being restarted if the requested music
class is the same as the one currently playing.
This includes an extra check to prevent the errors previously
experienced in the testsuite and has 100+ test runs behind it.
We have faced situation when using CDR and CEL by sqlite3 modules. With system having high load (~100 concurrent calls created by sipp) we found many cdr and cel records missed. There is special finction in sqlite3, that make able to fix this situation - sqlite3_wait_timeout, that also can replace awful code cdr_sqlite3 ad cel_sqlite3 modules. Also this function can be used for aastdb and res_config_sqlite3 to avoid missed writes to sqlite db.
#ASTERISK-23766 #close
Reported by: Igor Goncharovsky
Matthew Jordan [Mon, 16 Jun 2014 02:39:29 +0000 (02:39 +0000)]
channels/chan_sip: Forbid remote bridging if T.38 is negotiated
When a framehook is removed - such as the fax gateway framehook - the bridge
framework will re-evaluate the bridge mixing technologies to see if it can
improve the bridging. When this occurs, get_rtp_info will be called to
determine if local or remote bridging can be used. Using remote bridging
will cause a fax to fail, as direct media negotiation will cause some small
number of packets to not arrive at the remote endpoint.
This patch forces local native bridging if T.38 negotiation is in progress or
has been established.
Matthew Jordan [Sun, 15 Jun 2014 22:11:49 +0000 (22:11 +0000)]
channel_internal_api: Publish a snapshot change when linkedids change
Snapshots are now not published *quite* as much as they used to. One instance
where they are not published any longer is during bridge enter and exit - the
state of the channel doesn't change, the bridge does. However, channels are
changed when a linkedid is propagated; previously, the channel's state would
be updated and published during the bridge enter event. Now this must be
explicitly done.
We no longer publish a channel snapshot when it is associated with an endpoint;
after all, the channel itself hasn't changed - the endpoint state has changed.
This updates the channel_messages unit test accordingly.
Matthew Jordan [Sun, 15 Jun 2014 21:18:51 +0000 (21:18 +0000)]
MoH: Undo commit r416150 (1.8)
This patch reverts r416150. When the comparison between mohclass->name and
state->class->name is made, you are not guaranteed that (a) state->class is
non-NULL or that state or state->class are in a safe state.
Crashes caught by the bridges/transfer_capabilities test.
........
Merged revisions 416251 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........
Merged revisions 416252 from http://svn.asterisk.org/svn/asterisk/branches/11
Matthew Jordan [Fri, 13 Jun 2014 18:16:34 +0000 (18:16 +0000)]
stasis: Reduce creation of channel snapshots to improve performance
During some performance testing of Asterisk with AGI, ARI, and lots of Local
channels, we noticed that there's quite a hit in performance during channel
creation and releasing to the dialplan (ARI continue). After investigating
the performance spike that occurs during channel creation, we discovered
that we create a lot of channel snapshots that are technically unnecessary.
This includes creating snapshots during:
* AGI execution
* Returning objects for ARI commands
* During some Local channel operations
* During some dialling operations
* During variable setting
* During some bridging operations
And more.
This patch does the following:
- It removes a number of fields from channel snapshots. These fields were
rarely used, were expensive to have on the snapshot, and hurt performance.
This included formats, translation paths, Log Call ID, callgroup, pickup
group, and all channel variables. As a result, AMI Status,
"core show channel", "core show channelvar", and "pjsip show channel" were
modified to either hit the live channel or not show certain pieces of data.
While this is unfortunate, the performance gain from this patch is worth
the loss in behaviour.
- It adds a mechanism to publish a cached snapshot + blob. A large number of
publications were changed to use this, including:
- During Dial begin
- During Variable assignment (if no AMI variables are emitted - if AMI
variables are set, we have to make snapshots when a variable is changed)
- During channel pickup
- When a channel is put on hold/unhold
- When a DTMF digit is begun/ended
- When creating a bridge snapshot
- When an AOC event is raised
- During Local channel optimization/Local bridging
- When endpoint snapshots are generated
- All AGI events
- All ARI responses that return a channel
- Events in the AgentPool, MeetMe, and some in Queue
- Additionally, some extraneous channel snapshots were being made that were
unnecessary. These were removed.
- The result of ast_hashtab_hash_string is now cached in stasis_cache. This
reduces a large number of calls to ast_hashtab_hash_string, which reduced
the amount of time spent in this function in gprof by around 50%.
Kinsey Moore [Fri, 13 Jun 2014 13:10:15 +0000 (13:10 +0000)]
MoH: Don't restart stream on repeated start calls
Currently, music on hold will stop and then start again from the
beginning if ast_moh_start() is called multiple times. This can happen
if a call is put on hold repeatedly (the channel receives multiple
HOLD control frames) and can be triggered from ARI by starting MoH on a
channel multiple times. This is fairly jarring/annoying to users.
This change prevents MoH from being restarted if the requested music
class is the same as the one currently playing.
Corey Farrell [Thu, 12 Jun 2014 17:21:13 +0000 (17:21 +0000)]
Blocked revisions 415915
........
chan_sip: DEBUG messages in sdp_crypto.c display despite a DEBUG level of zero
Change debug level for messages in sdp_crypto.c from zero to one. This
ensures the messages are not displayed when debugging is disabled. Change
does not apply to 12+ as it was already fixed in those versions.
ASTERISK-23246 #close
Reported by: Rusty Newton
Review: https://reviewboard.asterisk.org/r/3605/
........
Merged revisions 415908 from http://svn.asterisk.org/svn/asterisk/branches/1.8
Richard Mudgett [Thu, 12 Jun 2014 16:41:56 +0000 (16:41 +0000)]
AST-2014-007: Fix DOS by consuming the number of allowed HTTP connections.
Simply establishing a TCP connection and never sending anything to the
configured HTTP port in http.conf will tie up a HTTP connection. Since
there is a maximum number of open HTTP sessions allowed at a time you can
block legitimate connections.
A similar problem exists if a HTTP request is started but never finished.
* Added http.conf session_inactivity timer option to close HTTP
connections that aren't doing anything. Defaults to 30000 ms.
* Removed the undocumented manager.conf block-sockets option. It
interferes with TCP/TLS inactivity timeouts.
* AMI and SIP TLS connections now have better authentication timeout
protection. Though I didn't remove the bizzare TLS timeout polling code
from chan_sip.
* chan_sip can now handle SSL certificate renegotiations in the middle of
a session. It couldn't do that before because the socket was non-blocking
and the SSL calls were not restarted as documented by the OpenSSL
documentation.
* Fixed an off nominal leak of the ssl struct in
handle_tcptls_connection() if the FILE stream failed to open and the SSL
certificate negotiations failed.
The patch creates a custom FILE stream handler to give the created FILE
streams inactivity timeout and timeout after a specific moment in time
capability. This approach eliminates the need for code using the FILE
stream to be redesigned to deal with the timeouts.
This patch indirectly fixes most of ASTERISK-18345 by fixing the usage of
the SSL_read/SSL_write operations.
ASTERISK-23673 #close
Reported by: Richard Mudgett
........
Merged revisions 415841 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........
Merged revisions 415854 from http://svn.asterisk.org/svn/asterisk/branches/11
app_queue: delayed state can cause early leavewhenempty ringing
In app_queue, device state changes arrive in event messages and
update the queue member status value. That value is checked in
get_member_status() to decide that the caller should leave when
there are no available members. Although event messages can be
delayed by other activity, there is no adverse affect by lagged
status except in one specific case: there is only one available
member, it was just rung, and leavewhenempty is enabled set for
ringing members. This change adds a direct check of the device
state only under this condition where the caller may be dropped
incorrectly, resolving this issue without affecting performance
of app_queue normally.
AST-1248 #close
Review: https://reviewboard.asterisk.org/r/3595/
Reported by: Thomas Arimont
........
Merged revisions 415833 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........
Merged revisions 415835 from http://svn.asterisk.org/svn/asterisk/branches/11
Jonathan Rose [Thu, 12 Jun 2014 15:26:23 +0000 (15:26 +0000)]
MixMontior: Add class authorization requirements to MixMonitor AMI commands
MixMonitor AMI commands StartMixMonitor and StopMixMonitor lacked class
authorization. StopMixMonitor now requires that the manager user either have
the call or system class authorization. StartMixMonitor is a slightly larger
issue since it can execute shell commands if the right arguments are passed
into it, and we consider this a permission escalation. A security release
will be issued for problem this shortly.
ASTERISK-23609 #close
Reported by: Corey Farrell
........
Merged revisions 415825 from http://svn.asterisk.org/svn/asterisk/branches/11
Kevin Harwell [Thu, 12 Jun 2014 14:38:34 +0000 (14:38 +0000)]
res_pjsip_pubsub: unauthenticated remote crash in PJSIP pub/sub framework
A remotely exploitable crash vulnerability exists in the PJSIP channel driver's
pub/sub framework. If an attempt is made to unsubscribe when not currently
subscribed and the endpoint's "sub_min_expiry" is set to zero, Asterisk tries
to create an expiration timer with zero seconds, which is not allowed, so an
assertion raised.
The fix was to reject a subscription that is attempting to unsubscribe when not
being already subscribed. Asterisk now checks for this situation appropriately
and responds with a 400 instead of crashing.
Mark Michelson [Thu, 12 Jun 2014 14:03:52 +0000 (14:03 +0000)]
Fix potential deadlock situation in res_pjsip.
SIP transaction timeouts are handled in the PJSIP monitor thread. When
this happens on a subscription, and the subscription is destroyed, the
subscription destruction is dispatched synchronously to the threadpool.
The issue is that the PJSIP dialog is locked by the monitor thread,
and then the dispatched task attempts to lock the dialog. This leads
to a deadlock that causes SIP traffic to no longer be accepted on the
Asterisk server.
The fix here is to treat the monitor thread as if it were a threadpool
thread when it attempts to dispatch synchronous tasks. This way, the
dispatched task turns into a simple function call within the same thread,
and the locking issue is averted.
Joshua Colp [Thu, 12 Jun 2014 11:33:13 +0000 (11:33 +0000)]
res_pjsip_pubsub: Persist subscriptions in sorcery so they are recreated on startup.
This change makes res_pjsip_pubsub persist inbound subscriptions in sorcery. By default
this uses the local astdb but it can also be configured to store within an outside
database. When Asterisk is started these subscriptions are recreated if they have not
expired. Notifications are sent to the devices which have subscribed and they are none
the wiser that the system has restarted.
Walter Doekes [Thu, 12 Jun 2014 07:47:45 +0000 (07:47 +0000)]
safe_asterisk: Overwrite old safe_asterisk on make install.
From now on, make install will overwrite safe_asterisk with the
latest version. You need to move any local modifications to files
inside /etc/asterisk/startup.d, if you have any.
See also commits r394939 and r397938.
ASTERISK-21965 #close
Patches:
safe_asterisk.patch uploaded by jkister (License 6232, modified by me)
Richard Mudgett [Wed, 11 Jun 2014 22:54:04 +0000 (22:54 +0000)]
format.c: Fix misuse of hash container function.
The supplied hash function to a container must be idempotent given the
object's key value to figure out which container bucket the object belongs
in. Returning a random number or the current container count is not
idempotent. The "computed hash" value doesn't help find the object later
in those cases.
* Fixed the format_list container to actually be a list since that is how
the container is used. Conceptually, if more than 283 formats were added
to the format_list then odd things may have happened before the fix.
........
Merged revisions 415728 from http://svn.asterisk.org/svn/asterisk/branches/11
Jonathan Rose [Tue, 10 Jun 2014 15:46:25 +0000 (15:46 +0000)]
PJSIP: PJSIPNotify - Strip content-length headers and add documentation
Documentation for how to add custom headers/content to notifies created
with the PJSIPNotify manager action was a little sparse and it also
wasn't vetting application of Content-length headers like its chan_sip
equivalent was (so two Content-length headers could be applied... and
PJSIP determines the content length anyway, so it just opens people up
for error). This patch also flips the variable order so that the
variables are interpreted in the same order as they are put in the AMI
action.
Mark Michelson [Mon, 9 Jun 2014 20:20:50 +0000 (20:20 +0000)]
chan_pjsip: Fix bug where custom SIP headers could be duplicated on outgoing INVITEs.
When using PJSIP_HEADER() to add custom headers to outgoing INVITE requests, certain
situations could result in the headers being duplicated. For instance, if the request
were retransmitted, or if the INVITE were re-sent with authentication credentials,
the custom headers would be re-added to the request.
The fix here is to, after adding the custom headers to the outbound INVITE, remove
the datastore that holds the custom headers to add. This way, there is no risk in
accidentally adding them if the session supplement is called into a second or third
time.
Walter Doekes [Mon, 9 Jun 2014 12:08:47 +0000 (12:08 +0000)]
safe_asterisk: Cleanup additions to r415132.
* Replaced a stray echo that should've been a message call in
safe_asterisk. This replaces a conditional log message by a slightly
different message. Please update your log parsing scripts.
* Made the $NOTIFY mail Subject more verbose by adding the machine name
and exitstatus.
(Note that a 'make install' still won't overwrite your old safe_asterisk
if it exists. See ASTERISK-21965.)
ASTERISK-23492 #close
........
Merged revisions 415521 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........
Merged revisions 415522 from http://svn.asterisk.org/svn/asterisk/branches/11
Corey Farrell [Mon, 9 Jun 2014 03:48:22 +0000 (03:48 +0000)]
autoservice: stop thread on graceful shutdown
This change adds thread shutdown to autoservice for graceful shutdowns only.
ast_register_cleanup is backported to 1.8 to allow this. The logger callid
is also released on shutdown in 11+.
Matthew Jordan [Sun, 8 Jun 2014 18:11:54 +0000 (18:11 +0000)]
bridges/bridge_native_rtp: Reconfigure bridge on removal of framehook
This patch is a re-do of r414122.
When r414122 was merged, a major problem with it was uncovered. UNBRIDGE soft
hangup flags have a catastrophic effect on the pbx core if they leak out from
the bridge layer: the channel gets hung up. With the number of threads
involved in a blind transfer, and with the initial patch, it was likely that
this would occur. This caused a large number of test failures
This patch is nearly identical with the one proposed in r414122, save for the
following changes:
- We explicitly clear the UNBRIDGE flag when setting an after goto on a
channel in a bridge
- Defensively, if we encounter an UNBRIDGE flag in the pbx core, we handle it
Kinsey Moore [Fri, 6 Jun 2014 19:08:42 +0000 (19:08 +0000)]
PJSIP: Remove premature write of raw formats
Currently, there are situations that can occur when using chan_pjsip
and certain dialplan applications (notably ChanSpy()) that can cause
the channel to get no audio with scrolling warnings about format
mismatches. This is caused by a failure to update translation paths on
a mid-call native format update since the raw formats have already
been updated by res_pjsip_sdp_rtp.c in set_caps(). Removing the
premature raw format updates allows the translation paths to be setup
correctly and the raw read and write formats with them.
Rusty Newton [Fri, 6 Jun 2014 12:48:33 +0000 (12:48 +0000)]
configs/cli_aliases.conf: Two new aliases, plus enhancements for context names.
Changed naming of included alias templates to avoid confusion between version names. For example, asterisk12 was for asterisk 1.2, so I changed it to asterisk_1dot2, so that later we can use asterisk_12 for Asterisk 12.
Added alias for "features reload" to the template for Asterisk 11 style syntax template, as features reload was removed in 12, but you can still do "module reload features"
Added alias for "pjsip reload" to the friendly template. It is shorter than "module reload res_pjsip.so" and if some are like me; I constantly forget that reloading chan_pjsip doesn't parse config. Remembering "pjsip reload" is just easier.
ASTERISK-23654 #close
ASTERISK-23654 #comment Fixed by adding two new aliases and enhancements for context names.
Richard Mudgett [Thu, 5 Jun 2014 17:51:28 +0000 (17:51 +0000)]
config: Fix config files not reloading when only an included file changes.
The twisted logic determining if a config file should be reloaded was
mostly broken and disabled. The incorrect test that ASTERISK-23383 fixed
actually reenabled the broken logic. The incorrect test was causing the
timestamp to always be cleared which caused config files with includes to
always be reloaded.
* Made wildcard includes always cause a reload. Determining if a file was
deleted cannot be determined without restructuring the cache to determine
if any files are missing from the last files actually loaded. Also
without refactoring config_text_file_load(), the glob loop couldn't check
more than one file for changes anyway.
* Made remove the cache entry if the file no longer exists when trying to
get its timestamp or it is no longer a regular file. This fixes the
corner case where the file was loaded, then deleted, then the config
reloaded, then the file restored with the same timestamp, and then the
config reloaded again.
* Made remove the cache entry include list when actually loading the file.
This gets rid of any stale includes the file had from the last time the
file was loaded.
Matthew Jordan [Thu, 5 Jun 2014 14:39:08 +0000 (14:39 +0000)]
app_confbridge: Allow muting of users waiting to enter a ConfBridge
Prior to this patch, users waiting to enter a ConfBridge were not considered
when muted via the CLI or via AMI. Instead, a confusing message would be
emitted stating that the channel did not exist.
This patch allows a user to be muted when waiting to enter a ConfBridge
conference. This is equivalent to start when muted, only toggled via the CLI
or AMI.
Review: https://reviewboard.asterisk.org/r/3582
#ASTERISK-23824 #close
patches:
rb3582.patch uploaded by tm1000 (License 6524)
........
Merged revisions 415206 from http://svn.asterisk.org/svn/asterisk/branches/11
Walter Doekes [Wed, 4 Jun 2014 20:14:16 +0000 (20:14 +0000)]
safe_asterisk: Cleanup and debian compatibility.
Cleans up the safe_asterisk script and adds the ASTSAFE_FOREGROUND
option that allows the debian asterisk init script to capture the
right pid.
* Drop the vim #modeline which wasn't used. Use test consistently
without the odd configure xno syntax. Double quote all paths.
General cleanup.
* Don't output message()s to the console but only to TTY if set.
* Allow TTY to be "no" as well as empty (debian compatibility with
debian/patches/safe_asterisk-config).
* Add option to export ASTSAFE_FOREGROUND=1 from the init script
that calls this to disable backgrounding. Debian uses a similar
method in debian/patches/safe_asterisk-nobg).
Matthew Jordan [Wed, 4 Jun 2014 14:09:59 +0000 (14:09 +0000)]
res_pjsip_session: Add debug statement for session refreshes
This small patch adds a debug level 3 statement indicating how a session
refresh is being sent - either as a re-INVITE or as an UPDATE - and where
the session refresh is going.
Corey Farrell [Wed, 4 Jun 2014 07:23:58 +0000 (07:23 +0000)]
app_confbridge: Correct verification of conference name length
Conference names were not checked for maximum length, allowing unexpected
behaviour. This change adds checking to ensure the maximum length is not
exceeded. The maximum length is also changed from 32 to AST_MAX_EXTENSION.
Walter Doekes [Tue, 3 Jun 2014 07:34:23 +0000 (07:34 +0000)]
func_odbc: Fix fixed size buffers fix (r414968).
The change that removed the fixed size buffers in odbc-related code --
removing arbitrary column width limits -- was incomplete. This change
adds: no segfault on writesql without insertsql and return value checks
after strdup.
While I was in the vicinity I cleaned up the linefeeds in the odbc
function descriptions, moved some code for clarity, removed some blobs
and noted (but didn't fix) that the 'odbc write ... exec' CLI command
doesn't behave as the dialplan equivalent when insertsql= is used.
Joshua Colp [Sun, 1 Jun 2014 15:31:22 +0000 (15:31 +0000)]
bridge_native_rtp: Take the bridge type choice of both channels into account.
The bridge_native_rtp module currently uses the bridge result of the first
channel that joins a bridge as the ultimate result. This means that if the
first channel has direct media enabled but the second does not a direct
media bridge will still occur.
This change makes it so that both sides are taken into account. If either
side forbids the bridge or responds with a local bridge result then
either a generic or local bridge occurs.
Matthew Jordan [Fri, 30 May 2014 12:39:36 +0000 (12:39 +0000)]
TALK_DETECT: A channel function that raises events when talking is detected
This patch adds a new channel function TALK_DETECT that, when set on a
channel, causes events indicating the start/stop of talking on a channel to be
emitted to both AMI and ARI clients.
The function allows setting both the silence threshold (the length of silence
after which we decide no one is talking) as well as the talking threshold (the
amount of energy that counts as talking). Parameters can be updated on a channel
after talk detection has been enabled, and talk detection can be removed at
any time.
The events raised by the function use a nomenclature similar to existing AMI/ARI
events.
For AMI: ChannelTalkingStart/ChannelTalkingStop
For ARI: ChannelTalkingStarted/ChannelTalkingFinished
Matthew Jordan [Fri, 30 May 2014 12:03:06 +0000 (12:03 +0000)]
main/config.c: AMI action UpdateConfig EmptyCat clears all categories
When invoking UpdateConfig AMI action with Action set to EmptyCat, Asterisk
will make all categories empty in the config but the one requested with a
Cat variable. This is due to a bug in ast_category_empty (main/config.c)
that makes an incorrect comparison for a category name.
This patch corrects the comparison such that only the requested category
is cleared.
Matthew Jordan [Wed, 28 May 2014 17:43:36 +0000 (17:43 +0000)]
ast-db-manage/cdr/env.py: Don't fail if a config file can't be loaded
When generating SQL files via the repotools alembic_creator.py script, a
configuration object is used programatically with SQLAlechemy, as opposed to
a configuration file. This patch ignores failures to interpret a config file,
as ... there isn't one in this case.
Richard Mudgett [Wed, 28 May 2014 16:54:31 +0000 (16:54 +0000)]
res_pjsip_session: Fix leaked video RTP ports.
Simply enabling PJSIP to negotiage a video codec (e.g., h264) would leak
video RTP ports if the codec were not negotiated by an incoming call.
* Made add_sdp_streams() associate the handler with the media stream if
the handler handled the media stream. Otherwise, when the
ast_sip_session_media object was destroyed it didn't know how to clean up
the RTP resources.
* Fixed sdp_requires_deferral() associating the handler with the media
stream when deciding if the SDP processing needs to be deferred for T.38.
Like the leaked video RTP ports, the T.38 handler needs to clean up
allocated resources from deciding if SDP processing needs to be deffered.
* Cleaned up some dead code in handle_incoming_sdp() and
sdp_requires_deferral().
Richard Mudgett [Wed, 28 May 2014 16:24:54 +0000 (16:24 +0000)]
app_agent_pool: Return to dialplan if the agent fails to ack the call.
Improvements to the agent pool functionality.
* AgentRequest no longer hangs up the caller if the agent fails to connect
with the caller. It now continues in the dialplan.
* AgentRequest returns AGENT_STATUS set to NOT_CONNECTED if the agent
failed to connect with the call. Most likely because the agent did not
acknowledge the call in time or got disconnected.
* The agent alerting play file configured by the agent.conf custom_beep
option can now be disabled by setting the option to an empty string. The
agent is effectively alerted to a call presence when MOH stops.
* Fixed bridge reference leak when the agent connects with a caller.
Walter Doekes [Tue, 27 May 2014 21:21:29 +0000 (21:21 +0000)]
chan_sip: Start session timer at 200, not at INVITE.
Asterisk started counting the session timer at INVITE while the other
end correctly started at 200. This meant that for short session-expiries
(90 seconds) combined with long ringing times (e.g. 30 seconds), asterisk
would wrongly assume that the timer was hit before the other end thought
it was time to send a session refresh. This resulted in prematurely
ended calls.
This changes the session timer to start counting first at 200 like RFC
says it should.
(Also removed a few excess NULL checks that would never hit, because if
they did, asterisk would have crashed already.)
Walter Doekes [Tue, 27 May 2014 20:02:06 +0000 (20:02 +0000)]
res_config_odbc: Fix old and new ast_string_field memory leaks.
The ODBC realtime driver uses ^NN parameter encoding to cope with the
special meaning of the semi-colon. A semi-colon in a field is
interpreted as if the key was supplied twice, something which isn't
otherwise possible with fixed database columns. E.g. allow=alaw;ulaw
is parsed as allow=alaw and allow=ulaw. A literal semi-colon is
rewritten to ^3B when stored in the database.
The module uses a stringfield to efficiently store the encoded
parameters. However, this stringfield wasn't always freed in some
off-nominal cases.
Commit r413241 fixed initialization so the encoding for INSERT and
DELETE queries wouldn't crash. (Only SELECTs and UPDATEs worked
apparently.) But that commit forgot the frees. This change cleans
that up.
Matthew Jordan [Sun, 25 May 2014 02:30:51 +0000 (02:30 +0000)]
core_unreal: Prevent double free of core_unreal pvt
When a channel is destroyed (such as via ast_channel_release in off nominal
paths in core_unreal), it will attempt to free (via ast_free) the channel tech
pvt. This is problematic for a few reasons:
1. The channel tech pvt is an ao2 object in core_unreal. Free'ing the pvt
directly is no good.
2. The channel tech pvt's reference count is dropped just prior to calling
ast_channel_release, resulting in the pvt's destruction. Hence, the
channel destructor is free'ing an invalid pointer.
This patch keeps the dropping of the reference count, but sets the pvt to
NULL on the channel prior to releasing it. This models what would occur if the
channel was hung up directly.
User events can now be generated from ARI. Events can be signalled with
arbitrary json variables, and include one or more of channel, bridge, or
endpoint snapshots. An application must be specified which will receive
the event message (other applications can subscribe to it). The message
will also be delivered via AMI provided a channel is attached. Dialplan
generated user event messages are still transmitted via the channel, and
will only be received by a stasis application they are attached to or if
the channel is subscribed to.
This change also introduces the multi object blob mechanism used to send
multiple snapshot types in a single message. The dialplan app UserEvent
was also changed to use multi object blob, and a new stasis message type
created to handle them.
Richard Mudgett [Thu, 22 May 2014 16:00:42 +0000 (16:00 +0000)]
app_meetme: Don't interrupt MOH for waitmarked users.
Occasionally, when the last marked user leaves the conference, waitmarked
users don't get MOH if MOH is supposed to be played while a waitmarked
user is waiting for another marked user.
* Made not interrupt MOH when the user is a waitmarked user. The
waitmarked user doesn't need to hear any leave announcements from the
conference as the user would have already heard different leave
announcements if they were enabled. Apparently DAHDI occasionally sends
unending non-silent streams to these users or a normal user still in the
conference has continuous high background noise. These non-silent streams
cause MOH to be suspended while the never ending "announcement" is played.
PJSIP would never send the final 200 Notify for a blind transfer
when transferring to parking. This patch fixes that. In addition,
it fixes a reference leak when performing blind transfers to
non-bridging extensions.
Matthew Jordan [Thu, 22 May 2014 11:51:56 +0000 (11:51 +0000)]
res_corosync: Update module to work with Stasis (and compile)
This patch fixes res_corosync such that it works with Asterisk 12. This
restores the functionality that was present in previous versions of
Asterisk, and ensures compatibility with those versions by restoring the
binary message format needed to pass information from/to them.
The following changes were made in the core to support this:
* The event system has been partially restored. All event definition and
event types in this patch were pulled from Asterisk 11. Previously, we had
hoped that this information would live in res_corosync; however, the
approach in this patch seems to be better for a few reasons:
(1) Theoretically, ast_events can be used by any module as a binary
representation of a Stasis message. Given the structure of an ast_event
object, that information has to live in the core to be used universally.
For example, defining the payload of a device state ast_event in
res_corosync could result in an incompatible device state representation
in another module.
(2) Much of this representation already lived in the core, and was not
easily extensible.
(3) The code already existed. :-)
* Stasis message types now have a message formatter that converts their
payload to an ast_event object.
* Stasis message forwarders now handle forwarding to themselves. Previously
this would result in an infinite recursive call. Now, this simply creates a
new forwarding object with no forwards set up (as it is the thing it is
forwarding to). This is advantageous for res_corosync, as returning NULL
would also imply an unrecoverable error. Returning a subscription in this
case allows for easier handling of message types that are published directly
to an aggregate topic that has forwarders.
Richard Mudgett [Wed, 21 May 2014 22:17:37 +0000 (22:17 +0000)]
core_unreal: Only block media frames when a generator is on both ends of an unreal channel.
The fix for ASTERISK-12292 was a bit too aggressive. You could have
generators pointed at each other on local channels but need to get other
kinds of frames such as DTMF or CONNECTED_LINE frames accross.
........
Merged revisions 414269 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........
Merged revisions 414270 from http://svn.asterisk.org/svn/asterisk/branches/11
pbx.c: prevent potential crash from recursive replace()
Recurisve usage of replace() resulted in corruption of the
temporary string storage and potential crash. By changing
the string to be allocated separtely per instance, this is
eliminated.
ASTERISK-23650 #comment Reported by: Roel van Meer
ASTEIRSK-23650 #close
Paul Belanger [Mon, 19 May 2014 19:50:44 +0000 (19:50 +0000)]
Replace __ast_answer with ast_raw_answer in app_control_answer
While load testing an ARI application, I noticed asterisk was returning HTTP 500
internal server errors on channels/:id/answer. After talking to #asterisk-dev,
the issue appeared to be a lack of media flowing after __ast_answer() was
called. So now, we call ast_raw_answer instead and no longer wait for media.
Matthew Jordan [Sun, 18 May 2014 20:29:12 +0000 (20:29 +0000)]
bridge_native_rtp/bridge_channel: Fix direct media issues due to frame hook
This patch fixes issues with direct media bridges that occur after a blind
transfer. These issues were caught by the (currently failing)
pjsip/transfers/blind_transfer/caller_direct_media test.
The test currently fails primarily for two reasons:
(1) When Bob and Charlie (the transfer target and the transfer destination)
enter a bridge together, the framehook remains on the transfer target
channel until both channels are in the bridge. As it consumes voice frames,
the initial bridge type is a simple bridge. The framehook is removed when
both channels are in the bridge; however, this does not currently cause the
bridging framework to re-evaluate the bridge. This patch adds a
AST_SOFTHANGUP_UNBRIDGE poke to the transfer target channel when a
framehook is removed so the bridge can re-evaluate itself.
(2) When a channel leaves a native RTP bridge, it may be leaving due to being
hung up. Sending a re-INVITE to a channel that is about to be hung up is
not nice - in fact, there's a good chance we'll send the BYE request before
the channel has had a chance to send back a 200 OK. To be somewhat nicer,
this patch adds a function to channel.h that allows the bridging framework
to query for exactly why a channel is leaving a bridge via the channel's
soft hangup flags. This allows it to only send the re-INVITE if there's a
chance the channel will survive the native bridging experience.
Richard Mudgett [Fri, 16 May 2014 17:30:10 +0000 (17:30 +0000)]
sig_pri.c: Pull the pri_dchannel() PRI_EVENT_RING case into its own function.
* Populate the CALLERID(ani2) value (and the special CALLINGANI2 channel
variable) with the ANI2 value in addition to the PRI specific ANI2 channel
variable.
* Made complete snapshot staging with the channel lock held. All channel
snapshots need to be done while the channel lock is held.
........
Merged revisions 414050 from http://svn.asterisk.org/svn/asterisk/branches/11
Richard Mudgett [Thu, 15 May 2014 21:54:06 +0000 (21:54 +0000)]
app_meetme: Fix overwrite of DAHDI conference data structure.
Starting a conference recording using the admin menu overwrites the DAHDI
conference data structure used to modify the admin user's conference mute
mode.
* Made no longer pass the user's DAHDI conference data structure into the
menu functions. The menu now uses its own DAHDI conference data
structure to start the recording channel.
* Moved the unlock conf->playlock to before playing the conf-full message.
No sense keeping the lock while that prompt is playing. The user is never
going to get into the conference at that point.
........
Merged revisions 413991 from http://svn.asterisk.org/svn/asterisk/branches/1.8
........
Merged revisions 413992 from http://svn.asterisk.org/svn/asterisk/branches/11
projects / thirdparty / asterisk.git / log
