This patch is related to a number of issues on the bug tracker that show
crashes related to freeing frames that came from a filestream. A number of
fixes have been made over time while trying to figure out these problems, but
there re still people seeing the crash. (Note that some of these bug reports
include information about other problems. I am specifically addressing
the filestream frame crash here.)
I'm still not clear on what the exact problem is. However, what is _very_
clear is that we have seen quite a few problems over time related to unexpected
behavior when we try to use embedded frames as an optimization. In some cases,
this optimization doesn't really provide much due to improvements made in other
areas.
In this case, the patch modifies filestream handling such that the embedded frame
will not be returned. ast_frisolate() is used to ensure that we end up with a
completely mallocd frame. In reality, though, we will not actually have to malloc
every time. For filestreams, the frame will almost always be allocated and freed
in the same thread. That means that the thread local frame cache will be used.
So, going this route doesn't hurt.
With this patch in place, some people have reported success in not seeing the
crash anymore.
(SWP-150)
(AST-208)
(ABE-1834)
(issue #15609)
Reported by: aragon
Patches:
filestream_frisolate-1.4.diff2.txt uploaded by russell (license 2)
Tested by: aragon, russell
(closes issue #15817)
Reported by: zerohalo
Tested by: zerohalo
(closes issue #15845)
Reported by: marhbere
Review: https://reviewboard.asterisk.org/r/386/
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@222878 65c4cc65-6c06-0410-ace0-fbb531ad65f3
See Mantis issue for details of what prompted this change.
Additional notes:
This patch changes the ao2_iterator API in two ways: F_AO2I_DONTLOCK
has become an enum instead of a macro, with a name that fits our
naming policy; also, it is now necessary to call
ao2_iterator_destroy() on any iterator that has been
created. Currently this only releases the reference to the container
being iterated, but in the future this could also release other
resources used by the iterator, if the iterator implementation changes
to use additional resources.
(closes issue #15987)
Reported by: kpfleming
Review: https://reviewboard.asterisk.org/r/383/
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@222152 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Be default, change SSRC when doing an audio stream changes Asterisk doesn't
honor marker bit when reinvited to already-bridged RTP streams,resulting in
far-end stack discarding packets with "old" timestamps that areactually part of
a new stream. This patch sends AST_CONTROL_SRCUPDATE whenever there is a
reinvite, unless the 'constantssrc' is set to true in sip.conf.
The original issue reported to Digium support detailed the following situation:
ITSP <-> Asterisk 1.4.26.2 <-> SIP-based Application Server Call comes in
fromITSP, Asterisk dials the app server which sends a re-invite back
toAsterisk--not to negotiate to send media directly to the ITSP, but to
indicatethat it's changing the stream it's sending to Asterisk. The app
servergenerates a new SSRC, sequence numbers, timestamps, and sets the marker
bit on the new stream. Asterisk passes through the teimstamp of the new stream,
butdoes not reset the SSRC, sequence numbers, or set the marker bit.
When the timestamp on the new stream is older than the timestamp on the
originalstream, the ITSP (which doesn't know there has been any change) discards
the newframes because it thinks they are too old. This patch addresses this by
changing the SSRC on a stream update unless constantssrc=true is set in
sip.conf.
Review: https://reviewboard.asterisk.org/r/374/
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@221086 65c4cc65-6c06-0410-ace0-fbb531ad65f3
This fix may potentially cause problems with CDR backends that access the channel a CDR is associated with via the channel list. This fix makes the channel unavabile at the time when the CDR backend is invoked. This has been documented in include/asterisk/cdr.h.
(closes issue #15316)
Reported by: vmarrone
Tested by: mnicholson
Review: https://reviewboard.asterisk.org/r/362/
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@219136 65c4cc65-6c06-0410-ace0-fbb531ad65f3
This fixes a few issues with incorrect extension states and adds
a cli command, core show device2extenstate, to display all possible
state mappings.
(closes issue #15413)
Reported by: legart
Patches:
exten_helper.diff uploaded by dvossel (license 671)
Tested by: dvossel, legart, amilcar
Review: https://reviewboard.asterisk.org/r/301/
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@204681 65c4cc65-6c06-0410-ace0-fbb531ad65f3
When the list to be appended is empty, and the list to be appended to is *not*,
AST_LIST_APPEND_LIST would actually cause the target list to become broken,
and no longer have a pointer to its last entry. This patch fixes the problem.
(reported by Stanislaw Pitucha on the asterisk-dev mailing list)
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@201261 65c4cc65-6c06-0410-ace0-fbb531ad65f3
There are various media paths in Asterisk (codec translators and UDPTL, primarily)
that can generate more than one frame to be generated when the application calling
them expects only a single frame. This patch addresses a number of those cases,
at least the primary ones to solve the known problems. In addition it removes the
broken TRACE_FRAMES support, fixes a number of bugs in various frame-related API
functions, and cleans up various code paths affected by these changes.
https://reviewboard.asterisk.org/r/175/
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@200991 65c4cc65-6c06-0410-ace0-fbb531ad65f3
We were setting the stack size for each thread to 240KB regardless of
architecture, which meant that in some scenarios we actually had less available
stack space on 64 bit processors (pointers use 8 bytes instead of 4). So now we
calculate the stack size we reserve based on the platform's __WORDSIZE, which
gives us:
32 bit -> 240KB
64 bit -> 496KB
128 bit -> 1008KB (that's right, we're ready for 128 bit processors)
Patch typed by me but written by several members of #asterisk-dev, including
Kevin, Tilghman, and Qwell.
(closes issue #14932)
Reported by: jpiszcz
Patches:
06052009_issue14932.patch uploaded by seanbright (license 71)
Tested by: seanbright
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@199626 65c4cc65-6c06-0410-ace0-fbb531ad65f3
During asterisk startup, a lock on the list of modules is obtained by the
primary thread while each module is initialized. Issue 13778 pointed out a
problem with this approach, however. Because the AMI is loaded before other
modules, it is possible for a module reload to be issued by a connected client
(via Action: Command), causing a deadlock.
The resolution for 13778 was to move initialization of the manager to happen
after the other modules had already been lodaded. While this fixed this
particular issue, it caused a problem for users (like FreePBX) who call AMI
scripts via an #exec in a configuration file (See issue 15189).
The solution I have come up with is to defer any reload requests that come in
until after the server is fully booted. When a call comes in to
ast_module_reload (from wherever) before we are fully booted, the request is
added to a queue of pending requests. Once we are done booting up, we then
execute these deferred requests in turn.
Note that I have tried to make this a bit more intelligent in that it will not
queue up more than 1 request for the same module to be reloaded, and if a
general reload request comes in ('module reload') the queue is flushed and we
only issue a single deferred reload for the entire system.
As for how this will impact existing installations - Before 13778, a reload
issued before module initialization was completed would result in a deadlock.
After 13778, you simply couldn't connect to the manager during startup (which
causes problems with #exec-that-calls-AMI configuration files). I believe this
is a good general purpose solution that won't negatively impact existing
installations.
(closes issue #15189)
(closes issue #13778)
Reported by: p_lindheimer
Patches:
06032009_15189_deferred_reloads.diff uploaded by seanbright (license 71)
Tested by: p_lindheimer, seanbright
Review: https://reviewboard.asterisk.org/r/272/
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@199022 65c4cc65-6c06-0410-ace0-fbb531ad65f3
The function ast_call_forward() forwards a call to an extension specified in an ast_channel's call_forward string. After an ast_channel is called, if the channel's call_forward string is set this function can be used to forward the call to a new channel and terminate the original one. I have included this api call in both channel.c's ast_request_and_dial() and res_feature.c's feature_request_and_dial(). App_dial and app_queue already contain call forward logic specific for their application and options.
(closes issue #13630)
Reported by: festr
Review: https://reviewboard.asterisk.org/r/271/
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@198891 65c4cc65-6c06-0410-ace0-fbb531ad65f3
This change also involves the addition of an AST_CDR_FLAG_ORIGINATED flag that is used on originated channels to distinguish: them from dialed channels.
(closes issue #12946)
Reported by: meral
Patches:
null-cdr2.diff uploaded by mnicholson (license 96)
Tested by: mnicholson, dbrooks
(closes issue #15122)
Reported by: sum
Tested by: sum
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@198068 65c4cc65-6c06-0410-ace0-fbb531ad65f3
When we receive a SIP reinvite, it is possible that we may not be able to process the
reinvite immediately since we have also sent a reinvite out ourselves. The problem is
that whoever sent us the reinvite may have also sent a reinvite out to another party,
and that reinvite may have succeeded.
As a result, even though we are not going to accept the reinvite we just received, it
is important for us to not have problems if we suddenly start receiving RTP from a new
source. The fix for this is to grab the media source information from the SDP of the
reinvite that we receive. This information is passed to the RTP layer so that it will
know about the alternate source for media.
Review: https://reviewboard.asterisk.org/r/252
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@197588 65c4cc65-6c06-0410-ace0-fbb531ad65f3
There are two flags being added to the chanspy audiohook here. One
is the pre-existing AST_AUDIOHOOK_TRIGGER_SYNC flag. With this set,
we ensure that the read and write slinfactories on the audiohook do
not skew beyond a certain tolerance.
In addition, there is a new audiohook flag added here,
AST_AUDIOHOOK_SMALL_QUEUE. With this flag set, we do not allow for
a slinfactory to build up a substantial amount of audio before
flushing it. For this particular issue, this means that the person
spying on the call will hear the conversations in real time with very
little delay in the audio.
(closes issue #13745)
Reported by: geoffs
Patches:
13745.patch uploaded by mmichelson (license 60)
Tested by: snblitz
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@197537 65c4cc65-6c06-0410-ace0-fbb531ad65f3
The bridge was terminating immediately after the attended transfer was
completed. The problem was because upon reentering ast_channel_bridge
nexteventts was checked to see if it was set and if so could possibly
return AST_BRIDGE_COMPLETE.
(closes issue #15183)
Reported by: andrebarbosa
Tested by: andrebarbosa, tootai, loloski
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@197124 65c4cc65-6c06-0410-ace0-fbb531ad65f3
This is accomplished by adding two functions to update the answer time and disposition of calls that checks for the proper lock flags. These functions are used in the ast_bridge_call() function so that ForkCDR(A) calls are respected.
This patch also modifies the way ast_bridge_call() chooses the cdr record to base the bridged_cdr on. Previously the first unlocked cdr record would be chosen, now instead the first cdr record is chosen and forked cdr records are moved to the bridge_cdr. This allows the original cdr record and any forked cdr records to be properly updated with answer and end times.
(closes issue #13797)
Reported by: sh0t
Tested by: sh0t
(closes issue #14744)
Reported by: deepesh
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@195881 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Attempt to make configure script regeneration 'safe' using autoconf 2.63, which embeds a bare CR into the script, thus making Subversion complain about inconsistent line endings
This commit changes the MIME type of the configure script to be 'binary' thus making Subversion no longer inspect line endings, and as a bonus 'svn diff' will no longer try to generate diff output for it, which is not generally useful anyway.
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@190721 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Initialized ast_call_feature in detect_disconnect to avoid accessing uninitialized memory. Cleaned up /param tags in features.h. No longer send dynamic features in ast_feature_detect.
issue #11583
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@183386 65c4cc65-6c06-0410-ace0-fbb531ad65f3
feature.conf has a disconnect option. By default this option is set to '*', but it could be anything. If a user wishes to disconnect a call before the other side answers, only '*' will work, regardless if the disconnect option is set to something else. This is because features are unavailable until bridging takes place. The default disconnect option, '*', was hardcoded in app_dial, which doesn't make any sense from a user perspective since they may expect it to be something different. This patch allows features to be detected from outside of the bridge, but not operated on. In this case, the disconnect feature can be detected before briding and handled outside of features.c.
(closes issue #11583)
Reported by: sobomax
Patches:
patch-apps__app_dial.c uploaded by sobomax (license 359)
11583.latest-patch uploaded by murf (license 17)
detect_disconnect.diff uploaded by dvossel (license 671)
Tested by: sobomax, dvossel
Review: http://reviewboard.digium.com/r/195/
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@183126 65c4cc65-6c06-0410-ace0-fbb531ad65f3
We have seen a number of problems caused by poll() not working properly on
Mac OSX. If you search around, you'll find a number of references to using
select() instead of poll() to work around these issues. In Asterisk, we've
had poll.c which implements poll() using select() internally. However, we
were still getting reports of problems.
vadim investigated a bit and realized that at least on his system, even
though we were compiling in poll.o, the system poll() was still being used.
So, the primary purpose of this patch is to ensure that we're using the
internal poll() when we want it to be used.
The changes are:
1) Remove logic for when internal poll should be used from the Makefile.
Instead, put it in the configure script. The logic in the configure
script is the same as it was in the Makefile. Ideally, we would have
a functionality test for the problem, but that's not actually possible,
since we would have to be able to run an application on the _target_
system to test poll() behavior.
2) Always include poll.o in the build, but it will be empty if AST_POLL_COMPAT
is not defined.
3) Change uses of poll() throughout the source tree to ast_poll(). I feel
that it is good practice to give the API call a new name when we are
changing its behavior and not using the system version directly in all cases.
So, normally, ast_poll() is just redefined to poll(). On systems where
AST_POLL_COMPAT is defined, ast_poll() is redefined to ast_internal_poll().
4) Change poll() in main/poll.c to be ast_internal_poll().
It's worth noting that any code that still uses poll() directly will work fine
(if they worked fine before). So, for example, out of tree modules that are
using poll() will not stop working or anything. However, for modules to work
properly on Mac OSX, ast_poll() needs to be used.
(closes issue #13404)
Reported by: agalbraith
Tested by: russell, vadim
http://reviewboard.digium.com/r/198/
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@182810 65c4cc65-6c06-0410-ace0-fbb531ad65f3
With these changes, for a module to export symbols into the global namespace, it must have *both* the AST_MODFLAG_GLOBAL_SYMBOLS flag and a linker script that allows the linker to leave the symbols exposed in the module's .so file (see res_odbc.exports for an example).
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@182808 65c4cc65-6c06-0410-ace0-fbb531ad65f3
With these changes, for a module to export symbols into the global namespace, it must have *both* the AST_MODFLAG_GLOBAL_SYMBOLS flag and a linker script that allows the linker to leave the symbols exposed in the module's .so file (see res_odbc.exports for an example).
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@182802 65c4cc65-6c06-0410-ace0-fbb531ad65f3
The main problem here was that cstdlib was undefining free thereby causing the
proper debug macros to not be used. ast_h323.cxx has been changed to call
ast_free instead to avoid the issue. Because using the ast prefix calls are
a better choice, ast_free_ptr is the new wrapper for free to pass to functions.
Also, a little bit of clean up was done to avoid the debug macros intentionally
being redefined.
(closes issue #13593)
Reported by: pj
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@181133 65c4cc65-6c06-0410-ace0-fbb531ad65f3
