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@182802 65c4cc65-6c06-0410-ace0-fbb531ad65f3
It is not safe to read the name field of an ast_channel without the channel
locked. This patch fixes some places in channel.c where this was being done,
and lead to crashes related to masquerades.
(closes issue #14623)
Reported by: guillecabeza
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@181423 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Since setting fdno to -1 had to be moved, a couple of other code paths that
do process an fd event return early and do not pass through the code path
where it was moved to. So, set it to -1 in a few other places, too.
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@179741 65c4cc65-6c06-0410-ace0-fbb531ad65f3
When you call ast_waitfor() on a channel, the index into the channel fds array
that holds the file descriptor that poll() determines has input available is
stored in fdno. This patch clears out this value after a call to ast_read()
and also reports errors if ast_read() is called without an fdno set.
From a discussion on the asterisk-dev list.
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@179608 65c4cc65-6c06-0410-ace0-fbb531ad65f3
This fixes a bad regression where the bridge would exit after an attended
transfer was made. The problem was due to nexteventts getting set after the
masquerade which caused the bridge to return AST_BRIDGE_COMPLETE.
(closes issue #14315)
Reported by: tim_ringenbach
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@179536 65c4cc65-6c06-0410-ace0-fbb531ad65f3
For example, with an IAX2 channel, you can have both the channel thread and the
chan_iax2 processing threads calling this function, and doing so twice at the
same time is a bad thing.
(Found in a debugging session with dvossel and mmichelson)
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@179461 65c4cc65-6c06-0410-ace0-fbb531ad65f3
The main problem is currently if the Dial flag L is used with a warning sound,
DTMF is not evaluated after the first warning sound. To fix this, a flag has
been added in ast_generic_bridge for playing the warning which ensures that if
a scheduled warning is missed, multiple warrnings are not played back (due to a
feature evaluation or waiting for digits). ast_channel_bridge was modified to
store the nexteventts in the ast_bridge_config structure as that information
was lost every time ast_channel_bridge was reentered, causing a hangup due to
incorrect time calculations.
(closes issue #14315)
Reported by: tim_ringenbach
Reviewed on reviewboard:
http://reviewboard.digium.com/r/163/
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@176701 65c4cc65-6c06-0410-ace0-fbb531ad65f3
There was a subtle change in ast_do_masquerade which
resulted in failed attempts to pickup calls. The problem
was that the value of the AST_FLAG_OUTGOING flag was
copied from the clone to the original channel. In the case
of call pickup, this meant that the AST_FLAG_OUTGOING flag
ended up being cleared on the channel that was attempting
to execute the pickup.
Because this flag was not set, when ast_read came across
an answer frame, it ignored it. The result of this was that
the calling channel was never properly answered.
This fix changes the behavior in ast_do_masquerade to set
the flags on the original channel to the union of the flags
on the clone channel. This way, if the AST_FLAG_OUTGOING
flag is set on either of the two channels involved in the
masquerade, the resulting channel will have the flag set
as well.
(closes issue #14206)
Reported by: francesco_r
Patches:
14206.patch uploaded by putnopvut (license 60)
Tested by: francesco_r, aragon, putnopvut
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@170392 65c4cc65-6c06-0410-ace0-fbb531ad65f3
The fix for this is to simply set the newly created datastore's data pointer
to NULL if it is inherited but has no duplicate callback.
(closes issue #14113)
Reported by: francesco_r
Patches:
14113.patch uploaded by putnopvut (license 60)
Tested by: francesco_r
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@166568 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
The issue that was reported was about a case where a RINGING channel got
redirected to an extension to pick up a call from parking. Once the parked
call got taken out of parking, it heard silence until the other side answered.
Ideally, the caller that was parked would get a ringing indication. This patch
fixes this case so that the caller receives ringback once it comes out of
parking until the other side answers.
The fixes are:
- Make sure we remember that a channel was an outgoing channel when doing
a masquerade. This prevents an erroneous ast_answer() call on the channel,
which causes a bogus 200 OK to be sent in the case of SIP.
- Add some additional comments to explain related parts of code.
- Update the handling of the ast_channel visible_indication field. Storing
values that are not stateful is pointless. Control frames that are events
or commands should be ignored.
- When a bridge first starts, check to see if the peer channel needs to be
given ringing indication because the calling side is still ringing.
- Rework ast_indicate_data() a bit for the sake of readability.
(closes issue #13747)
Reported by: davidw
Tested by: russell
Review: http://reviewboard.digium.com/r/90/
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@164201 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
Don't always define HAVE_ZAPTEL_CHANALARMS (since we check if it's defined..)
Minor cleanup to make things clear.
(closes issue #13726)
Reported by: tzafrir
Patches:
dahdi_def.diff uploaded by tzafrir (license 46)
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@150557 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Reported by: tomaso
Patches:
asterisk-1.6.0-rc2-cdrmemleak.patch uploaded by tomaso (license 564)
I basically spent the day, verifying that this patch
solves the problem, and doesn't hurt in non-problem
cases. Why valgrind did not plainly reveal this leak
absolutely mystifies and stuns me.
Many, many thanks to tomaso for finding and providing the fix.
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@140670 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
(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
less than the defined minimum digit length, and the other end only wants END
digits (SIP INFO, for example).
(closes issue #12778)
Reported by: tsearle
Patches:
12778.rev1.txt uploaded by russell (license 2)
Tested by: tsearle
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@121280 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
I ran into some problems with G.722 in 1.4, so I have merged in all of the fixes
in this area that I have made in trunk/1.6.0, and things are happy again.
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@114550 65c4cc65-6c06-0410-ace0-fbb531ad65f3
cleared an issue someone was seeing when attempting to show channels when
the load was high.
(closes issue #11667)
Reported by: falves11
Patches:
11677.txt uploaded by russell (license 2)
Tested by: falves11
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@114117 65c4cc65-6c06-0410-ace0-fbb531ad65f3
deadlock prevention in place in chan_local, but it would not work in a specific
case because the channel was recursively locked. By unlocking the channel prior
to calling the generator's generate callback in ast_read_generator_actions(), we
prevent the recursive locking, and therefore the deadlock.
(closes issue #12307)
Reported by: callguy
Patches:
12307.patch uploaded by putnopvut (license 60)
Tested by: callguy
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@113065 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
on the issue with mmichelson)
- Update copyright info on app_chanspy.
- Fix a race condition that caused app_chanspy to crash. The issue was that
the chanspy datastore magic that was used to ensure that spyee channels did
not disappear out from under the code did not completely solve the problem.
It was actually possible for chanspy to acquire a channel reference out of
its datastore to a channel that was in the middle of being destroyed. That
was because datastore destruction in ast_channel_free() was done near the
end. So, this left the code in app_chanspy accessing a channel that was
partially, or completely invalid because it was in the process of being free'd
by another thread. The following sort of shows the code path where the race
occurred:
=============================================================================
Thread 1 (PBX thread for spyee chan) || Thread 2 (chanspy)
--------------------------------------||-------------------------------------
ast_channel_free() ||
- remove channel from channel list ||
- lock/unlock the channel to ensure ||
that no references retrieved from ||
the channel list exist. ||
--------------------------------------||-------------------------------------
|| channel_spy()
- destroy some channel data || - Lock chanspy datastore
|| - Retrieve reference to channel
|| - lock channel
|| - Unlock chanspy datastore
--------------------------------------||-------------------------------------
- destroy channel datastores ||
- call chanspy datastore d'tor ||
which NULL's out the ds' || - Operate on the channel ...
reference to the channel ||
||
- free the channel ||
||
|| - unlock the channel
--------------------------------------||-------------------------------------
=============================================================================
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.4@108135 65c4cc65-6c06-0410-ace0-fbb531ad65f3
