These changes eliminate the need for (and use of)
the KEEPALIVE return code in res_features.c;
There are other places that use this result code
for similar purposes at a higher level, these appear
to be left alone in 1.4, but attacked in trunk.
The reason these changes are being made in 1.4, is
that parking ends a channel's life, in some situations,
and the code in the bridge (and some other places),
was not checking the result code properly, and dereferencing
the channel pointer, which could lead to memory corruption
and crashes.
Calling the masq_park function eliminates this danger
in higher levels.
A series of previous commits have replaced some parking calls
with masq_park, but this patch puts them ALL to rest,
(except one, purposely left alone because a masquerade
is done anyway), and gets rid of the code that tests
the KEEPALIVE result, and the NOHANGUP_PEER result codes.
While bug 13820 inspired this work, this patch does
not solve all the problems mentioned there.
I have tested this patch (again) to make sure I have
not introduced regressions.
Crashes that occurred when a parked party hung up
while the parking party was listening to the numbers
of the parking stall being assigned, is eliminated.
These are the cases where parking code may be activated:
1. Feature one touch (eg. *3)
2. Feature blind xfer to parking lot (eg ##700)
3. Run Park() app from dialplan (eg sip xfer to 700)
(eg. dahdi hookflash xfer to 700)
4. Run Park via manager.
The interesting testing cases for parking are:
I. A calls B, A parks B
a. B hangs up while A is getting the numbers announced.
b. B hangs up after A gets the announcement, but
before the parking time expires
c. B waits, time expires, A is redialed,
A answers, B and A are connected, after
which, B hangs up.
d. C picks up B while still in parking lot.
II. A calls B, B parks A
a. A hangs up while B is getting the numbers announced.
b. A hangs up after B gets the announcement, but
before the parking time expires
c. A waits, time expires, B is redialed,
B answers, A and B are connected, after
which, A hangs up.
d. C picks up A while still in parking lot.
Testing this throroughly involves acting all the permutations
of I and II, in situations 1,2,3, and 4.
Since I added a few more changes (ALL references to KEEPALIVE in the bridge
code eliimated (I missed one earlier), I retested
most of the above cases, and no crashes.
H-extension weirdness.
Current h-extension execution is not completely
correct for several of the cases.
For the case where A calls B, and A parks B, the
'h' exten is run on A's channel as soon as the park
is accomplished. This is expected behavior.
But when A calls B, and B parks A, this will be
current behavior:
After B parks A, B is hung up by the system, and
the 'h' (hangup) exten gets run, but the channel
mentioned will be a derivative of A's...
Thus, if A is DAHDI/1, and B is DAHDI/2,
the h-extension will be run on channel
Parked/DAHDI/1-1<ZOMBIE>, and the
start/answer/end info will be those
relating to Channel A.
And, in the case where A is reconnected to
B after the park time expires, when both parties
hang up after the joyful reunion, no h-exten
will be run at all.
In the case where C picks up A from the
parking lot, when either A or C hang up,
the h-exten will be run for the C channel.
CDR's are a separate issue, and not addressed
here.
As to WHY this strange behavior occurs,
the answer lies in the procedure followed
to accomplish handing over the channel
to the parking manager thread. This procedure
is called masquerading. In the process,
a duplicate copy of the channel is created,
and most of the active data is given to the
new copy. The original channel gets its name
changed to XXX<ZOMBIE> and keeps the PBX
information for the sake of the original
thread (preserving its role as a call
originator, if it had this role to begin
with), while the new channel is without
this info and becomes a call target (a
"peer").
In this case, the parking lot manager
thread is handed the new (masqueraded)
channel. It will not run an h-exten
on the channel if it hangs up while
in the parking lot. The h exten will
be run on the original channel instead,
in the original thread, after the bridge
completes.
See bug 13820 for our intentions as
to how to clean up the h exten behavior.
Review: http://reviewboard.digium.com/r/29/
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@166093 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Reported by: tzafrir
Replace a bunch of if defined checks for Zaptel/DAHDI through several new defines in dahdi_compat.h. This removes a lot of code duplication. Example from bug:
#ifdef HAVE_ZAPTEL
fd = open("/dev/zap/pseudo", O_RDWR);
#else
fd = open("/dev/dahdi/pseudo", O_RDWR);
#endif
is replaced with:
fd = open(DAHDI_FILE_PSEUDO, O_RDRW);
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@165991 65c4cc65-6c06-0410-ace0-fbb531ad65f3
One issue was that the ast_mutex_* API was being used within the context of the
thread local data destructors. We would go off and allocate more thread local data
while the pthread lib was in the middle of destroying it all. This led to a memory
leak.
Another issue was an invalid argument being provided to the the object_add
API call.
(closes issue #13678)
Reported by: ys
Tested by: russell
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@164736 65c4cc65-6c06-0410-ace0-fbb531ad65f3
These changes come from team/russell/issue_12658
1) Change autoservice to put digits on the head of the channel's frame readq
instead of the tail. If there were frames on the readq that autoservice
had not yet read, the previous code would have resulted in out of order
processing. This required a new API call to queue a frame to the head
of the queue instead of the tail.
2) Change up the processing of DTMF in ast_read(). Some of the problems
were the result of having two sources of pending DTMF frames. There
was the dtmfq and the more generic readq. Both were used for pending
DTMF in various scenarios. Simplifying things to only use the frame
readq avoids some of the problems.
3) Fix a bug where a DTMF END frame could get passed through when it
shouldn't have. If code set END_DTMF_ONLY in the middle of digit emulation,
and a digit arrived before emulation was complete, digits would get
processed out of order.
(closes issue #12658)
Reported by: dimas
Tested by: russell, file
Review: http://reviewboard.digium.com/r/85/
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@163448 65c4cc65-6c06-0410-ace0-fbb531ad65f3
parameters are not evaluated multiple times.
An example where this caused a problem was in chan_sip.c, with
the line
ast_string_field_set(p, fromdomain, ++fromdomain);
This patch was originally uploaded to issue #13783 by
jamessan. While the issue was closed for other reasons, this
patch is valid and fixes a separate problem, and is thus
being committed.
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@162670 65c4cc65-6c06-0410-ace0-fbb531ad65f3
The test is not valid. Besides, if we actually suspected that recursive
mutexes were not working, we would get a ton of LOG_ERROR messages when
DEBUG_THREADS is turned on.
(inspired by a discussion on the asterisk-dev list)
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@162413 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Reported by: ckjohnsonme
Patches:
14019.diff uploaded by murf (license 17)
Tested by: ckjohnsonme, murf
This crash was the result of a few small errors that
would combine in 64-bit land to result in a crash.
32-bit land might have seen these combine to mysteriously
drop the args to an application call, in certain
circumstances.
Also, in trying to find this bug, I spotted
a situation in the flex input, where, in passing
back a 'word' to the parser, it would allocate
a buffer larger than necessary. I changed the
usage in such situations, so that strdup was
not used, but rather, an ast_malloc, followed
by ast_copy_string.
I removed a field from the pval struct, in
u2, that was never getting used, and set in
one spot in the code. I believe it was an
artifact of a previous fix to make switch
cases work invisibly with extens.
And, for goto's I removed a '!' from
before a strcmp, that has been there
since the initial merging of AEL2, that
might prevent the proper target of a
goto from being found. This was pretty
harmless on its own, as it would just
louse up a consistency check for users.
Many thanks to ckjohnsonme for providing
a simplified and complete set of information
about the bug, that helped considerably in
finding and fixing the problem.
Now, to get aelparse up and running again
in trunk, and out of its "horribly broken" state,
so I can run the regression suite!
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@162013 65c4cc65-6c06-0410-ace0-fbb531ad65f3
since this branch already had some printf format attributes, enable checking for them and tag functions that didn't have them
format attributes in a consistent way
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@159808 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Reported by: tzafrir
When compiling against Zaptel dahdi_compat will now only define all the DAHDI defines if the Zaptel define is present. Also, there is no such thing as DAHDI_PRI.
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@158629 65c4cc65-6c06-0410-ace0-fbb531ad65f3
frame in a freed ast_filestream. This patch makes use of the
ao2 functions to make sure that we do not free an ast_filestream
structure until the embedded ast_frame has been "freed" as well.
(closes issue #13496)
Reported by: fst-onge
Patches:
filestream_frame_1_4.diff uploaded by putnopvut (license 60)
Tested by: putnopvut
Closes AST-89
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@158126 65c4cc65-6c06-0410-ace0-fbb531ad65f3
app_followme which would occur at the end of an attended
transfer. The error occurred because we initially stored
a pointer to an ast_channel which then was hung up due
to a masquerade.
This commit adds a "fixup" callback to the bridge_config
structure to allow for end_bridge_callback_data to be
changed in the case that a new channel pointer is needed
for the end_bridge_callback.
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@157305 65c4cc65-6c06-0410-ace0-fbb531ad65f3
try to find the channel by name in the list,
which is slow and resource consuming, but rather
to pay attention to the result codes from the
ast_bridge_call, to which I added the
AST_PBX_NO_HANGUP_PEER_PARKED value, which
now are returned when a channel is parked.
If you get AST_PBX_KEEPALIVE,
then don't touch the channel pointer.
If you get AST_PBX_NO_HANGUP_PEER, or
AST_PBX_NO_HANGUP_PEER_PARKED, then don't
touch the peer pointer.
Updated the several places where the results
from a bridge were not being properly obeyed,
and fixed some code I had introduced so that
the results of the bridge were not overridden
(in trunk).
All the places that previously tested for
AST_PBX_NO_HANGUP_PEER now have to check for
both AST_PBX_NO_HANGUP_PEER and AST_PBX_NO_HANGUP_PEER_PARKED.
I tested this against the 4 common parking
scenarios:
1. A calls B; B answers; A parks B; B hangs up while A is getting the parking
slot announcement, immediately after being put on hold.
2. A calls B; B answers; A parks B; B hangs up after A has been hung up, but
before the park times out.
3. A calls B; B answers; B parks A; A hangs up while B is getting the parking slot announcement, immediately after being put on hold.
4. A calls B; B answers; B parks A; A hangs up after B has been hung up, but before the park times out.
No crash.
I also ran the scenarios above against valgrind, and accesses looked good.
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@152535 65c4cc65-6c06-0410-ace0-fbb531ad65f3
so that if packetization of audio is close (but not equal)
we don't end up flushing the audiohooks over small
inconsistencies in synchronization.
Related to issue #13005, and solves the issue
for most people who were experiencing the problem.
However, a small number of people are still experiencing
the problem on long calls, so I am not closing
the issue yet
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@149204 65c4cc65-6c06-0410-ace0-fbb531ad65f3
- it is no longer necessary to forcibly include asterisk/autoconfig.h; every module already includes asterisk.h as its first header (even before system headers), which serves the same purpose
- astmm.h is now included by asterisk.h when needed, instead of being forced by the Makefile; this means external modules will build properly against installed headers with MALLOC_DEBUG enabled
- simplify the usage of some of these headers in the AEL-related stuff in the utils directory
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@144924 65c4cc65-6c06-0410-ace0-fbb531ad65f3
This is a "second attempt" to restore the previous "endbeforeh" behavior
in 1.4 and up. In order to capture information concerning all the
legs of transfers in all their infinite combinations, I was forced
to this particular solution by a chain of logical necessities, the
first being that I was not allowed to rewrite the CDR mechanism from
the ground up!
This change basically leaves the original machinery alone, which allows
IVR and local channel type situations to generate CDR's as normal, but
a channel flag can be set to suppress the normal running of the h exten.
That flag would be set by the code that runs the h exten from the
ast_bridge_call routine, to prevent the h exten from being run twice.
Also, a flag in the ast_bridge_config struct passed into ast_bridge_call
can be used to suppress the running of the h exten in that routine. This
would happen, for instance, if you use the 'g' option in the Dial app.
Running this routine 'early' allows not only the CDR() func to be used
in the h extension for reading CDR variables, but also allows them to
be modified before the CDR is posted to the backends.
While I dearly hope that this patch overcomes all problems, and
introduces no new problems, reality suggests that surely someone
will have problems. In this case, please re-open 13251 (or 13289),
and we'll see if we can't fix any remaining issues.
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@142675 65c4cc65-6c06-0410-ace0-fbb531ad65f3
The purpose of this branch was to take into account
"burps" which could cause jitterbuffers to misbehave.
One such example is if the L option to Dial() were used
to inject audio into a bridged conversation at regular
intervals. Since the audio here was not passed through
the jitterbuffer, it would cause a gap in the jitterbuffer's
timestamps which would cause a frames to be dropped for a
brief period.
Now ast_generic_bridge will empty and reset the jitterbuffer
each time it is called. This causes injected audio to be handled
properly.
ast_generic_bridge also will empty and reset the jitterbuffer
if it receives an AST_CONTROL_SRCUPDATE frame since the change
in audio source could negatively affect the jitterbuffer.
All of this was made possible by adding a new public API call
to the abstract_jb called ast_jb_empty_and_reset.
(closes issue #11259)
Reported by: plack
Tested by: putnopvut
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@135841 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Reported by: bcnit
Tested by: murf
I discovered that also, in the previous bug fixes and changes,
the cdr.conf 'unanswered' option is not being obeyed, so
I fixed this.
And, yes, there are two 'answer' times involved in this
scenario, and I would agree with you, that the first
answer time is the time that should appear in the CDR.
(the second 'answer' time is the time that the bridge
was begun).
I made the necessary adjustments, recording the first
answer time into the peer cdr, and then using that to
override the bridge cdr's value.
To get the 'unanswered' CDRs to appear, I purposely
output them, using the dial cmd to mark them as
DIALED (with a new flag), and outputting them if
they bear that flag, and you are in the right mode.
I also corrected one small mention of the Zap device
to equally consider the dahdi device.
I heavily tested 10-sec-wait macros in dial, and
without the macro call; I tested hangups while the
macro was running vs. letting the macro complete
and the bridge form. Looks OK. Removed all the
instrumentation and debug.
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@135799 65c4cc65-6c06-0410-ace0-fbb531ad65f3
app_chanspy should be set at load time, not at compile
time, since dahdi_chan_name is determined at load time.
Also changed the next_unique_id_to_use to have the
static qualifier.
Also added the dahdi_chan_name_len variable so that
strlen(dahdi_chan_name) isn't necessary. Thanks to
seanbright for the suggestion.
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@133169 65c4cc65-6c06-0410-ace0-fbb531ad65f3
correct registration of AMI actions in chan_dahdi; in zap-only mode, only register the Zap flavors of the actions (and use Zap prefixes for headers and acks), but in dahdi+zap mode, register both Zap and DAHDI flavors of actions
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@132787 65c4cc65-6c06-0410-ace0-fbb531ad65f3
and b) completes contexts correctly when the extension is ambiguous.
(closes issue #12980)
Reported by: licedey
Patches:
20080703__bug12980.diff.txt uploaded by Corydon76 (license 14)
Tested by: Corydon76
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@127973 65c4cc65-6c06-0410-ace0-fbb531ad65f3
(closes issue #10927)
Reported by: murf
Tested by: murf, deeperror
(closes issue #12907)
Reported by: falves11
Tested by: murf, falves11
(closes issue #11849)
Reported by: greyvoip
As to 11849, I think these changes fix the core problems
brought up in that bug, but perhaps not the more global
problems created by the limitations of CDR's themselves
not being oriented around transfers.
Reopen if necc, but bug reports are not the best
medium for enhancement discussions. We need to start
a second-generation CDR standardization effort to cover
transfers.
(closes issue #11093)
Reported by: rossbeer
Tested by: greyvoip, murf
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@127663 65c4cc65-6c06-0410-ace0-fbb531ad65f3
macro. This caused the lock to not actually be released, and as a result, not
avoid deadlocks at all. This resolves the issues reported in the last while about
Asterisk locking up all over the place (and most commonly, in chan_iax2).
(closes issue #12927)
(closes issue #12940)
(closes issue #12925)
(potentially closes others ...)
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@126573 65c4cc65-6c06-0410-ace0-fbb531ad65f3
don't include tonezone.h in dahdi_compat.h, because only a couple of modules need it
get app_rpt building again after the DAHDI changes
(closes issue #12911)
Reported by: tzafrir
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@125132 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Reported by: arkadia
Tested by: murf, arkadia
Options added to forkCDR() app and the CDR() func to
remove some roadblocks for CDR applications.
The "show application ForkCDR" output was upgraded
to more fully explain the inner workings of forkCDR.
The A option was added to forkCDR to force the
CDR system to NOT change the disposition on the
original CDR, after the fork. This involves
ast_cdr_answer, _busy, _failed, and so on.
The T option was added to forkCDR to force
obedience of the cdr LOCKED flag in the
ast_cdr_end, all the disposition changing
funcs (ast_cdr_answer, etc), and in the
ast_cdr_setvar func.
The CHANGES file was updated to explain ALL
the new options added to satisfy this bug report
(and some requests made verbally and via
email, irc, etc, over the past months/year)
The 's' option was added to the CDR() func,
to force it to skip LOCKED cdr's in the
chain.
Again, the new options should be totally transparent
to existing apps! Current behavior of CDR,
forkCDR, and the rest of the CDR system should
not change one little bit. Until you add the
new options, at least!
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@122046 65c4cc65-6c06-0410-ace0-fbb531ad65f3
After debugging a deadlock, it was noticed that when DEBUG_CHANNEL_LOCKS
is enabled in menuselect, the actual origin of channel locks is obscured
by the fact that all channel locks appear to happen in the function
ast_channel_lock(). This code change redefines ast_channel_lock to be a
macro which maps to __ast_channel_lock(), which then relays the proper
file name, line number, and function name information to the core lock
functions so that this information will be displayed in the case that
there is some sort of locking error or core show locks is issued.
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@116088 65c4cc65-6c06-0410-ace0-fbb531ad65f3
ignoring the way that macros expand. Instead, I have clarified in the
comment why the macro will work even if the scheduler id for the
task to be deleted changes during the execution of the macro.
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@115196 65c4cc65-6c06-0410-ace0-fbb531ad65f3
These changes address a critical performance issue introduced in the latest
release. The fix for the latest security issue included a change that made
Asterisk randomly choose call numbers to make them more difficult to guess by
attackers. However, due to some inefficient (this is by far, an understatement)
code, when Asterisk chose high call numbers, chan_iax2 became unusable after
just a small number of calls. On a small embedded platform, it would not be
able to handle a single call. On my Intel Core 2 Duo @ 2.33 GHz, I couldn't
run more than about 16 IAX2 channels. Ouch.
These changes address some performance issues of the find_callno() function
that have bothered me for a very long time. On every incoming media frame,
it iterated through every possible call number trying to find a matching
active call. This involved a mutex lock and unlock for each call number
checked. So, if the random call number chosen was 20000, then every media
frame would cause 20000 locks and unlocks. Previously, this problem was
not as obvious since Asterisk always chose the lowest call number it could.
A second container for IAX2 pvt structs has been added. It is an astobj2
hash table. When we know the remote side's call number, the pvt goes into
the hash table with a hash value of the remote side's call number. Then,
lookups for incoming media frames are a very fast hash lookup instead of an
absolutely insane array traversal.
In a quick test, I was able to get more than 3600% more IAX2 channels
on my machine with these changes.
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@114891 65c4cc65-6c06-0410-ace0-fbb531ad65f3
mansession_id cookie is coded to be limited to 8 characters of hex, and this
could break logins from 64-bit machines in some cases.
(inspired by AST-20)
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@114591 65c4cc65-6c06-0410-ace0-fbb531ad65f3
These changes make sure that the reference count for sip_peer objects properly
reflects the fact that the peer is sitting in the scheduler for a scheduled
callback for qualifying peers or for expiring registrations. Without this, it
was possible for these callbacks to happen at the same time that the peer was
being destroyed. This was especially likely to happen with realtime peers, and
for people making use of the realtime prune CLI command.
(closes issue #9520)
Reported by: kryptolus
Committed patch by me
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@114522 65c4cc65-6c06-0410-ace0-fbb531ad65f3
referenced, leading to memory corruption and eventual crashes. This code change ensures
that the dsp is freed when we are finished with the frame. This change is very similar
to a change Russell made with translators back a month or so ago.
(closes issue #11999)
Reported by: destiny6628
Patches:
11999.patch uploaded by putnopvut (license 60)
Tested by: destiny6628, victoryure
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@114207 65c4cc65-6c06-0410-ace0-fbb531ad65f3
datastore callback, called chan_fixup(). The concept is exactly like the
fixup callback that is used in the channel technology interface. This callback
gets called when the owning channel changes due to a masquerade. Before this
was introduced, if a masquerade happened on a channel being spyed on, the
channel pointer in the datastore became invalid.
(closes issue #12187)
(reported by, and lots of testing from atis)
(props to file for the help with ideas)
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@108583 65c4cc65-6c06-0410-ace0-fbb531ad65f3
it is appropriate and when it is not appropriate to use it.
I also removed the part of the debug message that mentions that this is probably a bug because
there are some perfectly legitimate places where ast_sched_del may fail to delete an entry (e.g.
when the scheduler callback manually reschedules with a new id instead of returning non-zero to
tell the scheduler to reschedule with the same idea). I also raised the debug level of the debug
message in AST_SCHED_DEL since it seems like it could come up quite frequently since the macro
is probably being used in several places where it shouldn't be. Also removed the redundant line,
file, and function information since that is provided by ast_log.
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@108227 65c4cc65-6c06-0410-ace0-fbb531ad65f3
and it is not worth spamming users with these messages unless we are pretty confident
that it should never happen. As it stands today, it _will_ and _does_ happen and
until that gets cleaned up a reasonable amount on the development side, let's not
spam the logs of everyone else.
(closes issue #12154)
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@106704 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Steve Murphy
Jeff Peeler
Russell Bryant
Mark Michelson
Joshua Colp
Sean Bright
Terry Wilson
Tilghman Lesher
Kevin P. Fleming
Brett Bryant