Richard Mudgett [Thu, 2 Jun 2016 23:19:13 +0000 (18:19 -0500)]
sorcery: Add setting object type congestion levels.
Sorcery creates taskprocessors for object types to process object observer
callbacks. An API call is needed to be able to set the congestion levels
of these taskprocessors for selected object types.
* Updated PJSIP's contact and contact_status sorcery object type observer
default congestion levels based upon stress testing. Increased the
congestion levels to reduce the potential for bursty register/unregister
and subscribe/unsubscribe activity from triggering the taskprocessor
overload alert.
Richard Mudgett [Thu, 2 Jun 2016 21:08:19 +0000 (16:08 -0500)]
taskprocessors: Implement high/low water mark alerts.
When taskprocessors get backed up, there is a good chance that we are
being overloaded and need to defer adding new work to the system.
* Implemented a high/low water alert mechanism for modules to check if the
system is being overloaded and take appropriate action. When a
taskprocessor is created it has default congestion levels set. A
taskprocessor can later have those congestion levels altered for specific
needs if stress testing shows that the taskprocessor is a symptom of
overloading or needs to handle bursty activity without triggering an
overload alert.
* Add CLI "core show taskprocessor" low/high water columns.
* Fixed __allocate_taskprocessor() to not use RAII_VAR(). RAII_VAR() was
never a good thing to use when creating a taskprocessor because of the
nature of how its references needed to be cleaned up on a partial
creation.
* Made res_pjsip's distributor check if the taskprocessor overload alert
is active before placing a message representing brand new work onto a
distributor serializer.
Richard Mudgett [Fri, 27 May 2016 22:31:52 +0000 (17:31 -0500)]
res_pjsip_session: Use distributor serializer for incoming calls.
We must continue using the serializer that the original INVITE came in on
for the dialog. There may be retransmissions already enqueued in the
original serializer that can result in reentrancy and message sequencing
problems.
Outgoing call legs create the pjsip/outsess/<endpoint> serializers for
their dialogs.
Richard Mudgett [Fri, 27 May 2016 17:50:14 +0000 (12:50 -0500)]
res_pjsip_pubsub.c: Use distributor serializer for incoming subscriptions.
We must continue using the serializer that the original SUBSCRIBE came in
on for the dialog. There may be retransmissions already enqueued in the
original serializer that can result in reentrancy and message sequencing
problems. The "sip_transaction Unable to register SUBSCRIBE transaction
(key exists)" message is a notable symptom of this issue.
Outgoing subscriptions still create the pjsip/pubsub/<endpoint>
serializers for their dialogs.
Richard Mudgett [Thu, 26 May 2016 22:35:04 +0000 (17:35 -0500)]
pjsip_distributor.c: Consistently pick a serializer for messages.
Incoming messages that are not part of a dialog or a recognized response
to one of our requests need to be sent to a consistent serializer. Under
load we may be queueing retransmissions before we can process the original
message. We don't need to throw these messages onto random serializers
and cause reentrancy and message sequencing problems.
* Created a pool of pjsip/distributor serializers that get picked by
hashing the call-id and remote tag strings of the received messages.
* Made ast_sip_destroy_distributor() destroy items in the reverse order of
creation.
Joshua Colp [Thu, 2 Jun 2016 17:04:45 +0000 (14:04 -0300)]
res_odbc: Implement a connection pool.
Testing has shown that our usage of UnixODBC is problematic
due to bugs within UnixODBC itself as well as the heavy weight
cost of connecting and disconnecting database connections, even
when pooling is enabled.
For users of UnixODBC 2.3.1 and earlier crashes would occur due
to insufficient protection of the disconnect operation. This was
fixed in UnixODBC 2.3.2 and above.
For users of UnixODBC 2.3.3 and higher a slow-down would occur
under heavy database use due to repeated connection establishment.
A regression is present where on each connection the database
configuration is cached again, with the cache growing out of
control.
The connection pool implementation present in this change helps
to mitigate these issues by reducing how much we connect and
disconnect database connections. We also solve the issue of
crashes under UnixODBC 2.3.1 by defaulting the maximum number of
connections to 1, returning us to the previous working behavior.
For users who may have a fixed version the maximum concurrent
connection limit can be increased helping with performance.
The connection pool works by keeping a list of active connections.
If the connection limit has not been reached a new connection is
established. If the connection limit has been reached then the
request waits until a connection becomes available before
continuing.
Joshua Colp [Thu, 2 Jun 2016 09:59:06 +0000 (06:59 -0300)]
alembic: Fix migration.
The 81b01a191a46_pjsip_add_contact_reg_server.py script was attempting
to use UniqueConstraint and failing. It was not imported and after
importing it also continued to fail.
I've changed the script to use the explicit name of the constraint
instead.
Richard Mudgett [Wed, 1 Jun 2016 18:57:53 +0000 (13:57 -0500)]
logging,cdr,cel: Fix stringfield memory leak.
The stringfields refactor to allow adding stringfields to the end of a
structure (f6f4cf459f43f072604927209b39646f84aaa2e2) exposed some
incomplete cleanup code by some stringfield users.
The most noticeable leaker is the logging system where there is a leak for
every log message generated.
ASTERISK-26078 #close
Reported by: Etienne Lessard
Patches:
jira_asterisk_26078_v13.patch (license #5621) patch uploaded
by Richard Mudgett
Richard Mudgett [Tue, 31 May 2016 18:02:15 +0000 (13:02 -0500)]
pjsip_distributor.c: Use correct rdata info access method (Part 2).
The pjproject doxygen for rdata->msg_info.info says to call
pjsip_rx_data_get_info() instead of accessing the struct member directly.
You need to call the function mostly because the function will generate
the struct member value if it is not already setup.
George Joseph [Mon, 30 May 2016 15:58:35 +0000 (09:58 -0600)]
pjproject_bundled: Move to pjproject 2.5
Although all the patches we had against 2.4.5 were applied by Teluu,
a new bug was introduced preventing re-use of tcp and tls transports
This patch removes all the previous patches against 2.4.5, updates
the version to 2.5, and adds a new patch to correct the transport
re-use problem.
Richard Mudgett [Wed, 25 May 2016 23:30:07 +0000 (18:30 -0500)]
pjsip_distributor.c: Use correct rdata info access method.
The pjproject doxygen for rdata->msg_info.info says to call
pjsip_rx_data_get_info() instead of accessing the struct member directly.
You need to call the function mostly because the function will generate
the struct member value if it is not already setup.
Alexei Gradinari [Fri, 20 May 2016 18:56:30 +0000 (14:56 -0400)]
app_voicemail: fix bugs, imap mm_status log change to debug
Fixed some bugs:
- create dirpath when save downloading message from IMAP storage.
- create IMAP folder if not exists when saving to IMAP storage
- check if file successfully opened before write to it
- some IMAP checks
- remove non-standard flag 'Unseen'
etc
Change to debug IMAP mm_status log instead of verbose.
Remove unused X-Asterisk-VM-Caller-channel message header
for security reason. The clients should not know name of peer/endpoint.
Alexei Gradinari [Thu, 19 May 2016 19:56:26 +0000 (15:56 -0400)]
res_pjsip: add "via_addr", "via_port", "call_id" to contact
As res_pjsip_nat rewrites contact's address, only the last Via header
can contain the source address of registered endpoint.
Also Call-Id header may contain the source address of registered
endpoint.
Added "via_addr", "via_port", "call_id" to contact.
Added new fields ViaAddress, CallID to AMI event ContactStatus.
Alexei Gradinari [Tue, 24 May 2016 21:56:49 +0000 (17:56 -0400)]
res_pjsip: chatty verbose messages
There are a lot of verbose messages about Endpoint and Contact status
changes if there are many dynamic endpoints.
The patch sets verbose level 2 for Endpoint status changes
and verbose level 3 for Contact status changes.
Corey Farrell [Fri, 12 Feb 2016 15:59:44 +0000 (10:59 -0500)]
threadpool: Fix potential data race.
worker_start checked for ZOMBIE status without holding a lock. All
other read/write of worker status are performed with a lock, so this
check should do the same.
Alexei Gradinari [Tue, 10 May 2016 19:30:29 +0000 (15:30 -0400)]
func_odbc: single database connection should be optional
func_odbc was changed in Asterisk 13.9.0
to make func_odbc use a single database connection per DSN
because of reported bug ASTERISK-25938
with MySQL/MariaDB LAST_INSERT_ID().
This is drawback in performance when func_odbc is used
very often in dialplan.
Mark Michelson [Fri, 20 May 2016 14:39:10 +0000 (09:39 -0500)]
res_pjsip: Match dialogs on responses better.
When receiving an incoming response to a dialog-starting INVITE, we were
not matching the response to the INVITE dialog. Since we had not
recorded the to-tag to the dialog structure, the PJSIP-provided method
to find the dialog did not match.
Most of the time, this was not a problem, because there is a fall-back
that makes the response get routed to the same serializer that the
request was sent on. However, in cases where an asynchronous DNS lookup
occurs in the PJSIP core, the thread that sends the INVITE is not
actually a threadpool serializer thread. This means we are unable to
record a serializer to handle the incoming response.
Now, imagine what happens when an INVITE is sent on a non-serialized
thread, and an error response (such as a 486) arrives. The 486 ends up
getting put on some random threadpool thread. Eventually, a hangup task
gets queued on the INVITE dialog serializer. Since the 486 is being
handled on a different thread, the hangup task can execute at the same
time that the 486 is being handled. The hangup task assumes that it is
the sole owner of the INVITE session and channel, so it ends up
potentially freeing the channel and NULLing the session's channel
pointer. The thread handling the 486 can crash as a result.
This change has the incoming response match the INVITE transaction, and
then get the dialog from that transaction. It's the same method we had
been using for matching incoming CANCEL requests. By doing this, we get
the INVITE dialog and can ensure that the 486 response ends up being
handled by the same thread as the hangup, ensuring that the hangup runs
after the 486 has been completely handled.
Joshua Colp [Thu, 19 May 2016 16:41:45 +0000 (13:41 -0300)]
res_sorcery_astdb: Filter fields to only the registered ones.
This change introduces the same filtering that is done in res_sorcery_realtime
to the res_sorcery_astdb module. This allows persisted sorcery objects
that may contain unknown fields to still be read in from the AstDB
and used. This is particularly useful when switching between different
versions of Asterisk that may have introduced additional fields.
snuffy [Tue, 10 May 2016 02:40:08 +0000 (12:40 +1000)]
res_pjsip_empty_info: Respond to empty SIP INFO packets
Some SBCs require responses to empty SIP INFO packets
after establishing call via INVITE, if not responded to
they may drop your call after unspecified timeout of X minutes.
They are identified by having no Content-Type, check for this
and respond with 200 - OK message.
* Local fax starts rtp call to remote fax
* Remote fax starts t38 call back to local fax.
* Local fax sends t38 no-signal to Asterisk before sending an OK.
* udptl processes the frame and increments the expected sequence number.
* chan_sip drops the frame because the call isn't up so nothing goes out
the external interface to open the port for incoming packets.
* Local fax sends OK and Asterisk sends OK to the remote fax.
* Remote fax sends t38 packets which are dropped by the firewall.
* Local fax re-sends t38 no-signal with the same sequence number.
* udptl drops the frame because it thinks it's a dup.
* Still no outgoing packets to open the firewall.
* t38 negotiation fails.
The patch drops frames t38 received before udptl sequence processing
when the call hasn't been answered yet. The second no-signal frame
is then seen as new and is relayed out the external interface which
opens the port and allows negotiation to continue.
George Joseph [Tue, 17 May 2016 16:14:51 +0000 (10:14 -0600)]
chan_sip: Prevent extra Session-Expires headers from being added
When chan_sip does a re-INVITE to refresh a session and authentication
is required, the INVITE with the Authorization header containes a
second Session-Expires header without the ";refersher=" parameter.
This is causing some proxies to return a 400. Also, when Asterisk is
the uas and the refresher, it is including the Session-Expires and
Min-SE headers in OPTIONS messages which is not allowed per RFC4028.
This patch (based on the reporter's) Checks to see if a Session-Expires
header is already in the message before adding another one. It also
checks that the method is INVITE or UPDATE.
George Joseph [Mon, 16 May 2016 20:29:38 +0000 (14:29 -0600)]
res_pjsip_outbound_registration: Clean up state when registration is deleted
Nothing was cleaning up the registration state object when ast_sorcery_delete
was called on a registration. So, the registration was deleted from sorcery
but the state object went right on refreshing the registration (or failing
to refresh the registration) with the peer.
* Added a 'deleted' observer on registration that removes the state object.
George Joseph [Mon, 16 May 2016 00:05:34 +0000 (18:05 -0600)]
res_pjsip: Set TCP_NODELAY on TCP transports
Although it's perfectly legal to place multiple SIP messages in the same packet,
it can cause problems because the Linux default is to enable Path MTU Discovery
which sets the Don't Fragment bit on the packets. If adding a second message to
the packet causes the MTU to be exceeded, and the destination isn't equipped to
send a FRAGMENTATION NEEDED response to a large packet, the packet will just be
dropped.
We can't specifically tell the stack to send only 1 message per packet, but we
can turn on TCP_NODELAY when we create the transport. This will at least tell
the stack to send packets as soon as possible.
ASTERISK-26005 #close Reported-by: Ross Beer
Change-Id: I820f23227183f2416ca5e393bec510e8fe1c8fbd
Matt Jordan [Thu, 12 May 2016 12:08:08 +0000 (07:08 -0500)]
res/res_hep_pjsip: Fix reported local IP address when bound to 'any'
When bound to an 'any' address, e.g., 0.0.0.0, PJSIP reports as its
local address the 'any' address, as opposed to the IP address we
actually received the packet on. This can cause some confusion in Homer,
as it will dutifully report what we send it.
This patch uses the PJSIP inspection routines to determine which IP
address we probably received the packet on based on the remote party's
IP address. In the event that this fails, it falls back to the IP
address natively reported by the transport.
Alexei Gradinari [Fri, 13 May 2016 16:38:20 +0000 (12:38 -0400)]
res_pjsip: Endpoint IP Access Controls
With the old SIP module we can use IP access controls per peer.
PJSIP module missing this feature.
This patch added next configuration Endpoint options:
"acl" - list of IP ACL section names in acl.conf
"deny" - List of IP addresses to deny access from
"permit" - List of IP addresses to permit access from
"contact_acl" - List of Contact ACL section names in acl.conf
"contact_deny" - List of Contact header addresses to deny
"contact_permit" - List of Contact header addresses to permit
This patch also better logging failed request:
add custom message instead of "No matching endpoint found"
add SIP method to logging
Matt Jordan [Thu, 12 May 2016 01:17:15 +0000 (20:17 -0500)]
res_hep: Provide an option to pick the UUID type
At one point in time, it seemed like a good idea to use the Asterisk
channel name as the HEP correlation UUID. In particular, it felt like
this would be a useful identifier to tie PJSIP messages and RTCP
messages together, along with whatever other data we may eventually send
to Homer. This also had the benefit of keeping the correlation UUID
channel technology agnostic.
In practice, it isn't as useful as hoped, for two reasons:
1) The first INVITE request received doesn't have a channel. As a
result, there is always an 'odd message out', leading it to be
potentially uncorrelated in Homer.
2) Other systems sending capture packets (Kamailio) use the SIP Call-ID.
This causes RTCP information to be uncorrelated to the SIP message
traffic seen by those capture nodes.
In order to support both (in case someone is trying to use res_hep_rtcp
with a non-PJSIP channel), this patch adds a new option, uuid_type, with
two valid values - 'call-id' and 'channel'. The uuid_type option is used
by a module to determine the preferred UUID type. When available, that
source of a correlation UUID is used; when not, the more readily available
source is used.
For res_hep_pjsip:
- uuid_type = call-id: the module uses the SIP Call-ID header value
- uuid_type = channel: the module uses the channel name if available,
falling back to SIP Call-ID if not
For res_hep_rtcp:
- uuid_type = call-id: the module uses the SIP Call-ID header if the
channel type is PJSIP and we have a channel,
falling back to the Stasis event provided
channel name if not
- uuid_type = channel: the module uses the channel name
Tzafrir Cohen [Tue, 10 May 2016 08:08:33 +0000 (11:08 +0300)]
basic-cfg: asterisk.conf: debug level 5 spams
Don't suggest users to use debug level 5, which spews (usually
non-useful) debug information. Reduce the suggestion to (an
arbitrarily-selected) level 2.
Tzafrir Cohen [Tue, 10 May 2016 08:10:55 +0000 (11:10 +0300)]
basic-cfg: asterisk.conf: don't set languages
* No need to set language in a miniml configuration. 'en' will do just
fine.
* It would be useful to have an example of setting it to a different
language.
* Setting the documentation language explicitly is likewise not
required. Setting it to a different value is not common. At least
until there is a set of translated documentation.
Tzafrir Cohen [Tue, 10 May 2016 13:17:29 +0000 (16:17 +0300)]
followme: delete the right recorded name file
FollowMe with the option a records the name of the caller and plays it
to the callee. However it has failed to clean up that recorded file
as it tried to delete the file name without the '.sln' extension.
Mark Michelson [Thu, 12 May 2016 19:36:25 +0000 (14:36 -0500)]
Use doubles instead of floats for conversions when comparing strings.
In 13.9.0, there was an issue where PJSIP contacts added to an AOR would
be deleted at seemingly random times.
One reason this was happening was because of an operation to retrieve
the contacts whose expiration time was less than or equal to the current
time. When retrieving existing contacts, the contact's expiration time
and the current time were converted from a string to a float, and those
two floats were compared.
On some systems, including mine, this conversion was horribly off. For
instance, I could regularly see the string "1463079214" get converted
into 1463079168.000000. When switching from using a float to using a
double, the conversion was as expected.
Why was the conversion to float off? My best guess is that the
conversion to float was attempting to store the entire value in the 23
bit significand of the IEEE-754 floating point number. In particular, if
you take only the 23 most significant bits of 1463079214, you get the
messed up 1463079168 that we were seeing in the conversion. It likely
was possible to get a more precise value by composing the number using
an exponent, but the conversion did not work that way. With a double,
you have a 52 bit significand, allowing the entire value to fit there,
and thereby allowing an accurate conversion.
Sebastian Damm [Tue, 10 May 2016 15:19:48 +0000 (17:19 +0200)]
res_pjsip_outbound_registration: generate correct Contact URI for TLS
There are two types of SIP URIs indicating a secure transport:
* sips:user@example.org
* sip:user@example.org;transport=tls
When using a sips URI, Asterisk checks incoming INVITEs and answers from
the other side for sips URIs, and rejects the packet if there are only
sip URIs. So Asterisk should only generate a sips Contact URI if the
other side supports it.
This patch makes Asterisk generate either a sip or sips Contact URI
depending on the format of the server URI.
If you want a sip URI, use:
server_uri=sip:example.org\;transport=tls
If you want a sips URI, use:
server_uri=sips:example.org
ASTERISK-25990 #close Reported-by: Sebastian Damm
Change-Id: I5ae57d6531ce940b5fc64d5cd2673e60db0f9ba2
Matt Jordan [Wed, 11 May 2016 19:07:17 +0000 (14:07 -0500)]
configure: Fix errors with AST_UNDEFINED_SANITIZER/AST_LEAK_SANITIZER
When running on a system that does not support or use AST_UNDEFINED_SANITIZER
or AST_LEAK_SANITIZER, the configure script would incorrectly set those
constants to a blank value, e.g., 'AST_UNDEFINED_SANITIZER='. This would
cause menuselect to error out, complaining that a blank value is not a
valid option. This patch corrects the issue by setting the value to 0 if
the options that those constants enable/disable is not found.
Kevin Harwell [Tue, 3 May 2016 20:43:16 +0000 (15:43 -0500)]
res_pjsip_outbound_publish: state potential dropped on reloads/realtime fetches
When reloading, or fetching realtime data, if the "apply" failed for any
numerous reasons the current state object would not be maintained. This
potentially resulted in publishes being stopped for some states/clients when
they should not have been.
This patch makes it so the current state object is kept upon any type of reload/
fetch failures.
Kevin Harwell [Tue, 3 May 2016 20:31:19 +0000 (15:31 -0500)]
res_pjsip_outbound_publish: Potential crash due to off nominal path
It was possible for the explicit publish destroy function to be called without
the pjsip client ever being initialized. This fix checks to make sure there is
a client to destroy before attempting.
Kevin Harwell [Tue, 3 May 2016 20:35:24 +0000 (15:35 -0500)]
res_pjsip_outbound_publishing: After unloading the library won't load again
The same thing was happening in res_pjsip_publish_asterisk. When the library
was unloaded it did not unregister the object type from sorcery. Subsequent
loads resulted in a failed load due to the sorcery type already existing.
Kevin Harwell [Tue, 3 May 2016 20:39:32 +0000 (15:39 -0500)]
res_pjsip_outbound_publish: Won't unload if condition wait times out
When res_pjsip_outbound_publish unloads it has to wait for all current
publishing objects to get done. However if the wait condition times out
then it does not fail the unload. This sometimes results in an infinite
loop check while unloading. This patch now fails the unload operation if
the condition times out.
Kevin Harwell [Thu, 5 May 2016 16:37:37 +0000 (11:37 -0500)]
res_pjsip_authenticator_digest: Don't use source port in nonce verification
From the issue reporter:
"res_pjsip_outbound_authenticator_digest builds a nonce that is a hash of
the timestamp, the source address, the source port, a server UUID that is
calculated at startup, and the authentication realm.
Rather than caching nonces that we create, we instead attempt to re-calculate
the nonce when receiving an incoming request with authentication. We then
compare the re-calculated nonce to the incoming nonce, and if they don't match,
then authentication has failed early.
The problem is that it is possible, especially when using TCP, to receive two
requests from the same endpoint but have differing source ports for those
requests. Asterisk itself commonly will use different source ports for
outbound TCP requests."
This patch removes the source port dependency when building the nonce.
George Joseph [Sat, 7 May 2016 19:39:25 +0000 (13:39 -0600)]
config_transport: Tell pjproject to allow all SSL/TLS protocols
The default tls settings for pjproject only allow TLS 1, TLS 1.1 and TLS 1.2.
SSL is not allowed. So, even if you specify "sslv3" for a transport method,
it's silently ignored and one of the TLS protocols is used. This was a new
behavior of pjsip_tls_setting_default() in 2.4 (when tls.proto was added) that
we never caught.
Now we need to set tls.proto = 0 after we call pjsip_tls_setting_default().
This tells pjproject to set the socket protocol to match the method.
Jaco Kroon [Wed, 4 May 2016 07:40:55 +0000 (09:40 +0200)]
app_confbridge: Add a regcontext option for confbridge bridge profiles.
This patch allows for having app_confbridge register the name of the
conference as an extension into a specific context, similar to
regcontext for chan_sip. This variant is not quite as involved as the
one in chan_sip and doesn't allow for multiple contexts or custom
extensions, you can only specify the context and the conference name
will always be used as the extension to register.
George Joseph [Mon, 9 May 2016 01:19:50 +0000 (19:19 -0600)]
pjproject_bundled: Check for python-dev and TEST_FRAMEWORK
The pjsua and pjsystest apps are now built only if TEST_FRAMEWORK is set.
The python bindings are now built only if TEST_FRAMEWORK is set and a
python development package is installed.
projects / thirdparty / asterisk.git / log
