Due to a faulty function for debugging reference decrementing, it was possible
to reduce the refcount on the wrong object if two moh classes of the same name
were in the moh class container.
(closes issue ASTERISK-22252)
Reported by: Walter Doekes
Patches:
18_moh_debug_ref_patch.diff Uploaded by Jonathan Rose (license 6182)
........
Merged revisions 398937 from http://svn.asterisk.org/svn/asterisk/branches/1.8
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@421777 65c4cc65-6c06-0410-ace0-fbb531ad65f3
This change fixes up DTLS support in res_rtp_asterisk so it can accept and provide
a SHA-256 fingerprint, so it occurs on RTCP, and so it occurs after ICE negotiation
completes. Configuration options to chan_sip have also been added to allow behavior
to be tweaked (such as forcing the AVP type media transports in SDP).
ASTERISK-22961 #close
Reported by: Jay Jideliov
Review: https://reviewboard.asterisk.org/r/3679/
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@417677 65c4cc65-6c06-0410-ace0-fbb531ad65f3
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.
ASTERISK-23917 #close
Reported by: Matt Jordan
Review: https://reviewboard.asterisk.org/r/3624/
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@417310 65c4cc65-6c06-0410-ace0-fbb531ad65f3
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.
Review: https://reviewboard.asterisk.org/r/3615/
........
Merged revisions 416439 from http://svn.asterisk.org/svn/asterisk/branches/1.8
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@416440 65c4cc65-6c06-0410-ace0-fbb531ad65f3
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
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@416252 65c4cc65-6c06-0410-ace0-fbb531ad65f3
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.
Review: https://reviewboard.asterisk.org/r/3615/
........
Merged revisions 416150 from http://svn.asterisk.org/svn/asterisk/branches/1.8
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@416151 65c4cc65-6c06-0410-ace0-fbb531ad65f3
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
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@415854 65c4cc65-6c06-0410-ace0-fbb531ad65f3
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.
Review: https://reviewboard.asterisk.org/r/3555/
........
Merged revisions 414564 from http://svn.asterisk.org/svn/asterisk/branches/1.8
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@414565 65c4cc65-6c06-0410-ace0-fbb531ad65f3
The realtime API specifies that the store callback is supposed to return the number
of rows affected. res_config_pgsql was instead returning an Oid cast as an int, which
during any nominal execution would be cast to 0. Returning 0 when more than 0 rows were
inserted causes problems to the function's callers.
To give an idea of how strange code can be, this is the necessary code change to fix
a device state issue reported against chan_pjsip in Asterisk 12+. The issue was that
the registrar would attempt to insert contacts into the database. Because of the 0
return from res_config_pgsql, the registrar would think that the contact was not successfully
inserted, even though it actually was. As such, even though the contact was query-able
and it was possible to call the endpoint, Asterisk would "think" the endpoint was unregistered,
meaning it would report the device state as UNAVAILABLE instead of NOT_INUSE.
The necessary fix applies to all versions of Asterisk, so even though the bug reported
only applies to Asterisk 12+, the code correction is being inserted into 1.8+.
Closes issue ASTERISK-23707
Reported by Mark Michelson
........
Merged revisions 413224 from http://svn.asterisk.org/svn/asterisk/branches/1.8
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@413225 65c4cc65-6c06-0410-ace0-fbb531ad65f3
This resolves a race condition where data could be written to a NULL
FILE pointer causing a crash as a websocket connection was in the
process of shutting down by adding locking to websocket session writes
and by deferring session teardown until session destruction.
(closes issue ASTERISK-23605)
Review: https://reviewboard.asterisk.org/r/3481/
Reported by: Matt Jordan
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@413123 65c4cc65-6c06-0410-ace0-fbb531ad65f3
On congested networks, it is possible for the DTLS handshake messages to get
lost. This patch adds a timer to res_rtp_asterisk that will periodically
check to see if the handshake has succeeded. If not, it will retransmit the
DTLS handshake.
Review: https://reviewboard.asterisk.org/r/3337
ASTERISK-23649 #close
Reported by: Nitesh Bansal
patches:
dtls_retransmission.patch uploaded by Nitesh Bansal (License 6418)
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@413008 65c4cc65-6c06-0410-ace0-fbb531ad65f3
This patch fixes setting nullable integer columns to NULL instead of an empty
string, which fails for PostgreSQL, for example. The current code is supposed
to do so, but the check is broken. The patch also allows the first column in
the list to be a nullable integer.
This patch also adds a compatibility setting in res_odbc.conf,
allow_empty_string_in_nontext. It is enabled by default. It should be disabled
for database backends (such as PostgreSQL) that require NULL instead of an
empty string for Integer columns.
Review: https://reviewboard.asterisk.org/r/3375
(issue ASTERISK-23459)
Reported by: zvision
patches:
res_config_odbc.diff uploaded by zvision (License 5755)
........
Merged revisions 411399 from http://svn.asterisk.org/svn/asterisk/branches/1.8
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@411408 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Per Johann Steinwendtner on the asterisk-dev mailing list:
http://lists.digium.com/pipermail/asterisk-dev/2014-March/066102.html
g711_free() was introduced in spandsp 0.0.6pre4 and g711_release() became a
noop. I opted not to remove the call to g711_release() since it is harmless
and to call g711_free() if we have a sufficiently recent version of spandsp.
(issue ASTERISK-20149)
Reported by: Alexandr Gordeev
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@410829 65c4cc65-6c06-0410-ace0-fbb531ad65f3
I observed a crash in res_musiconhold on an Asterisk 11 system using realtime
MOH. Investigation of the backtrace showed a corrupt mohclass, implying that
it got destroyed before the code expected it to. I went looking for reference
counting errors that could have caused this crash and this patch this result.
It contains 2 changes.
1) Remove a usless block of code that was impossible to reach. There was even
a comment indicating that it was impossible to reach. The conditional includes
"!ast_test_flag(global_flags, MOH_CACHERTCLASSES)" and it's inside of an if
block with the opposite check "ast_test_flag(global_flags,
MOH_CACHERTCLASSES)". There's no good reason to keep it around.
2) A similar block to #1 contained a reference counting error. It stores
state->class in the local variable mohclass without increasing its reference
count. The reference count on mohclass is decremented at the end of the
function. This block of code probably very rarely runs, which would help
explain why this system was working fine for many months before experiencing a
crash.
Review: https://reviewboard.asterisk.org/r/3282/
........
Merged revisions 410043 from http://svn.asterisk.org/svn/asterisk/branches/1.8
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@410044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
When acting as a T.38 fax gateway, res_fax_spandsp would at times cause a crash
in libspandsp. This would occur when, during fax tone detection, a ulaw/alaw
frame would be passed to modem_connect_tones_rx. That particular routine
expects the data to be in slin format. This patch looks at the frame type and,
if the data is ulaw/alaw, converts the format to slin before passing it to
modem_connect_tones_rx.
Review: https://reviewboard.asterisk.org/r/3296
(closes issue ASTERISK-20149)
Reported by: Alexandr Gordeev
Tested by: Michal Rybarik
patches:
spandsp_g711decode.diff uploaded by Michal Rybarik (license 6578)
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@409990 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Several fixes for the WebSockets implementation in res/res_http_websocket.c
* Flush the websocket session FILE* as fwrite() may not actually guarantee sending
the data to the network. If we do not flush, it seems that buffering on the SSL
socket for outbound messages causes issues
* Refactored ast_websocket_read to take into account that SSL file descriptors
may be ready to read via fread() but poll() will not actually say so because
the data was already read from the network buffers and is now in the libc buffers
(closes issue ASTERISK-23099)
(closes issue ASTERISK-21930)
Review: https://reviewboard.asterisk.org/r/3248/
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@409681 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Prior to this patch, local candidate lists including SRFLX would fail to start
properly when building ICE candidate check lists. This patch fixes that problem
by making sure that each SRFLX candidate is associated with the proper
base address so that the check list can create matches properly.
This patch was written by jcolp. The issue will be left open to await testing
by the issue participants.
(issue ASTERISK-23213)
Reported by: Andrea Suisani
Review: https://reviewboard.asterisk.org/r/3256/
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@409129 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Change minrate from 2400 to 4800 on config reload in response to changes from
ASTERISK-22790 only. Any config with minrate of 2400 that would fail before
r405693 will still fail.
Comment out many settings in res_fax.conf.sample. The defaults are set in
res_fax.c, so setting the same value in sample config does nothing but make
the sample config more fragile.
(closes issue ASTERISK-23231)
Reported by: David Brillert
Review: https://reviewboard.asterisk.org/r/3261/
........
Merged revisions 409052 from http://svn.asterisk.org/svn/asterisk/branches/1.8
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@409053 65c4cc65-6c06-0410-ace0-fbb531ad65f3
The code assumed that unregistering the alias would always succeed while in
practice this is not actually true. A common case is the "reload" command itself.
If the cli_aliases.conf configuration file was changed and reload executed the
command would fail to unregister and ultimately point to freed memory.
The reload process now checks whether unregistering succeeded or not and if not
the old CLI alias is retained.
(closes issue ASTERISK-19773)
Reported by: Joel Vandal
(closes issue ASTERISK-22757)
Reported by: Gareth Blades
........
Merged revisions 407205 from http://svn.asterisk.org/svn/asterisk/branches/1.8
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@407210 65c4cc65-6c06-0410-ace0-fbb531ad65f3
ast_bind to a port reserved for another program by SELinux causes
errno == EACCES. This caused random failures when binding rtp or
udptl sockets. Treat EACCES as a non-fatal error, try next port.
(closes issue ASTERISK-23134)
Reported by: Corey Farrell
........
Merged revisions 406933 from http://svn.asterisk.org/svn/asterisk/branches/1.8
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@406934 65c4cc65-6c06-0410-ace0-fbb531ad65f3
This restricts direct usage of global oseq so that all accesses are
locked and threads are not racing to get oseq values that they did not
claim.
This also fixes a build error in res_pktccops under dev mode.
(closes issue ASTERISK-23100)
Reported by: adomjan
Patch by: adomjan
........
Merged revisions 406037 from http://svn.asterisk.org/svn/asterisk/branches/1.8
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@406038 65c4cc65-6c06-0410-ace0-fbb531ad65f3
According to the new standard for V.27 and V.32 they are able to transmit
at a bit rate of 4,800 or 9,600. The check_mode_rate function needed to be
updated to reflect this. Also, because of this change the default 'minrate'
value was updated to be 4800.
(closes issue ASTERISK-22790)
Reported by: Paolo Compagnini
Patches:
res_fax.txt uploaded by looserouting (license 6548)
........
Merged revisions 405656 from http://svn.asterisk.org/svn/asterisk/branches/1.8
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@405693 65c4cc65-6c06-0410-ace0-fbb531ad65f3
A 'make distclean' is equivalent to 'make dist-clean' in the top most Makefile.
This patch updates the res/Makefile to recognize both distclean and dist-clean.
Note that this is needed for removing build.mak, which can run into problems
if the source file of Asterisk or its path is changed after build.mak is
generated.
(issue ASTERISK-22480)
Reported by: Matt Jordan
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@405362 65c4cc65-6c06-0410-ace0-fbb531ad65f3
In ast_rtp_ice_start if the ice session create check list failed, start check
was never initiated and ice_started was never set to true. Upon re-entering
the function (for instance, [un]hold) it would try to create the check list
again with duplicate remote candidates.
Fixed so that if the create check list fails the necessary data structures
are properly re-initialized for any subsequent retries.
Note, it was decided to not stop ice support (by calling ast_rtp_ice_stop) on a
check list failure because it possible things might still work. However, a
debug message was added to help with any future troubleshooting.
(closes issue ASTERISK-22911)
Reported by: Vytis Valentinavičius
Patches:
works_on_my_machine.patch uploaded by xytis (license 6558)
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@405234 65c4cc65-6c06-0410-ace0-fbb531ad65f3
In fax_detect_framehook() a null pointer reference can occur where a
voice frame is processed but no dsp is attached to the fax detection
structure. The code block that rejects frames that detection cannot
be processed on is checking for dsp but falls through when it should
instead return, as this change implements.
(closes issue ASTERISK-22942)
Reported by: adomjan
Review: https://reviewboard.asterisk.org/r/3076/
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@404351 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Prior to this patch, res_fax_spandsp was conservative with how it initialized
the spandsp T.38 context. It would only initialize it if the driver thought
the current state was a T.38 fax. While this works fine in nominal situations,
in certain off nominal situations, res_fax_spandsp can believe that a T.38
fax will not occur when in fact one has started. In particular, this was
discovered when res_fax would fall back to audio after timing out on a T.38
upgrade. The SIP channel driver would continue to retry the re-INVITE and -
if the remote end responded after res_fax timed out with a 200 OK - a T.38
frame would be delivered to the res_fax stack when it no longer expected it.
As it turns out, there does not appear to be any downside to always
initializing the T.38 context, other than the actual memory allocation.
Since that avoids this off nominal situation (and others which are equally
likely hard to predict), this is the safest way to avoid this problem.
Much thanks to Torrey as well for providing a scenario that reproduces this
issue.
(closes issue ASTERISK-21242)
Reported by: Ashley Winters
Tested by: Torrey Searle
patches:
always-init-t38.patch uploaded by awinters (License 6477)
A_PARTY.xml uploaded by tsearle (License 5334)
........
Merged revisions 403449 from http://svn.asterisk.org/svn/asterisk/branches/1.8
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@403450 65c4cc65-6c06-0410-ace0-fbb531ad65f3
This corrects one-way audio between Asterisk and Chrome/jssip as a
result of Asterisk inserting the incorrect RTCP port into RTCP SRFLX
ICE candidates. This also exposes an ICE component enumeration to
extract further details from candidates.
(closes issue ASTERISK-21383)
Reported by: Shaun Clark
Review: https://reviewboard.asterisk.org/r/2967/
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@402345 65c4cc65-6c06-0410-ace0-fbb531ad65f3
In r400089, a patch was put in to correct erroneous RTCP statistic resets.
Unfortunately, ast_rtp_read can be called on an RTP instance that does not
have RTCP information. This patch prevents that crash by only resetting
the statistics if we do actually have an RTCP instance.
(issue AST-1174)
(closes issue ASTERISK-22667)
Reported by: John Bigelow
........
Merged revisions 401445 from http://svn.asterisk.org/svn/asterisk/branches/1.8
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/11@401446 65c4cc65-6c06-0410-ace0-fbb531ad65f3
