Russell Bryant [Tue, 3 Mar 2009 16:48:50 +0000 (16:48 +0000)]
Merged revisions 179742 via svnmerge from
https://origsvn.digium.com/svn/asterisk/trunk
................
r179742 | russell | 2009-03-03 10:47:28 -0600 (Tue, 03 Mar 2009) | 14 lines
Merged revisions 179741 via svnmerge from
https://origsvn.digium.com/svn/asterisk/branches/1.4
........
r179741 | russell | 2009-03-03 10:45:46 -0600 (Tue, 03 Mar 2009) | 6 lines
Ensure chan->fdno always gets reset to -1 after handling a channel fd event.
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.
Move where fdno is set to the default value to *after* the read callback of the channel driver is called.
We have to do this as the underlying channel driver may need the fdno value to determine what to read.
........
................
Russell Bryant [Tue, 3 Mar 2009 13:56:06 +0000 (13:56 +0000)]
Merged revisions 179609 via svnmerge from
https://origsvn.digium.com/svn/asterisk/trunk
................
r179609 | russell | 2009-03-03 07:54:41 -0600 (Tue, 03 Mar 2009) | 17 lines
Merged revisions 179608 via svnmerge from
https://origsvn.digium.com/svn/asterisk/branches/1.4
........
r179608 | russell | 2009-03-03 07:53:52 -0600 (Tue, 03 Mar 2009) | 9 lines
Make it easier to detect an improper call to ast_read().
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.
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.
Russell Bryant [Mon, 2 Mar 2009 23:39:56 +0000 (23:39 +0000)]
Merged revisions 179533 via svnmerge from
https://origsvn.digium.com/svn/asterisk/trunk
................
r179533 | russell | 2009-03-02 17:36:38 -0600 (Mon, 02 Mar 2009) | 48 lines
Merged revisions 179532 via svnmerge from
https://origsvn.digium.com/svn/asterisk/branches/1.4
........
r179532 | russell | 2009-03-02 17:34:13 -0600 (Mon, 02 Mar 2009) | 40 lines
Move ast_waitfor() down to avoid the results of the API call becoming stale.
This call to ast_waitfor() was being done way too soon in this section of code.
Specifically, there was code in between the call to waitfor and the code that
uses the result that puts the channel in autoservice. By putting the channel
in autoservice, the previous results of ast_waitfor() become meaningless,
as the autoservice thread will do it's own ast_waitfor() and ast_read()
on the channel.
So, when we came back out of autoservice and eventually hit the block of code
that calls ast_read() on the channel, there may not actually be any input on
the channel available. Even though the previous call to ast_waitfor() in
app_meetme said there was input, the autoservice thread has since serviced
the channel for some period of time.
This bug manifested itself while dvossel was doing some testing of MeetMe in
Asterisk trunk. He was using the timerfd timing module. When the code hit
ast_read() erroneously, it determined that it must have been called because of
input on the timer fd, as chan->fdno was set to AST_TIMING_FD, since that was
the cause of the last legitimate call to ast_read() done by autoservice.
In this test, an IAX2 channel was calling into the MeetMe conference. It was
_much_ more likely to be seen with an IAX2 channel because of the way audio
is handled. Every audio frame that comes in results in a call to
ast_queue_frame(), which then uses ast_timer_enable_continuous() to notify
the channel thread that a frame is waiting to be handled. So, the chances
of ast_waitfor() indicating that a channel needs servicing due to a timer
event on an IAX2 event is very high.
Finally, it is interesting to note that if a different timing interface was
being used, this bug would probably not be noticed. When ast_read() is called
and erroneously thinks that there is a timer event to handle, it calls the
ast_timer_ack() function. The pthread and dahdi timing modules handle the
ack() function being called when there is no event by simply ignoring it.
In the case of the timerfd module, it results in a read() on the timer fd
that will block forever, as there is no data to read. This caused Asterisk
to lock up very quickly.
Thanks to dvossel and mmichelson for the fun debugging session. :-)
When ending a recording with silence detection, remember to reduce the duration.
The end of the recording is correspondingly trimmed, but the duration was not
trimmed by the number of seconds trimmed, so the saved duration was necessarily
longer than the actual soundfile duration.
(closes issue #14406)
Reported by: sasargen
Patches:
20090226__bug14406.diff.txt uploaded by tilghman (license 14)
Tested by: sasargen
........
................
Russell Bryant [Mon, 2 Mar 2009 23:04:18 +0000 (23:04 +0000)]
Merged revisions 179462 via svnmerge from
https://origsvn.digium.com/svn/asterisk/trunk
................
r179462 | russell | 2009-03-02 17:00:30 -0600 (Mon, 02 Mar 2009) | 16 lines
Merged revisions 179461 via svnmerge from
https://origsvn.digium.com/svn/asterisk/branches/1.4
........
r179461 | russell | 2009-03-02 16:58:18 -0600 (Mon, 02 Mar 2009) | 8 lines
Ensure that only one thread is calling ast_settimeout() on a channel at a time.
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)
Properly free memory and remove scheduler entries when a transmission failure occurs.
Previously, only the "data" field of the sip_pkt created during __sip_reliable_xmit
was freed when XMIT_ERROR was returned by __sip_xmit. When retrans_pkt was called,
this inevitably resulted in the reading and writing of freed memory.
XMIT_ERROR is a condition meaning that we don't want to attempt resending the packet
at all. The proper action to take is to remove the scheduler entry we just created,
free the packet's data as well as the packet itself, and unlink it from the list of
packets on the sip_pvt structure.
Special thanks goes to DEA who had to redo this patch twice
because we first put unload/load support in and later redid the way
we configure devices and lines.
(closes issue #10297)
Reported by: DEA
Patches:
skinny-reload-trunkv2.diff uploaded by wedhorn (license 30)
skinny-reload-trunk-v4.txt uploaded by DEA (license 3)
With mods by me based on feedback from wedhorn and Russell and seanbright
Tested by: DEA, mvanbaak, pj
This change moves the default feature digit timeout to 1000 ms from the previous default of 500.
As per bug 14515, a dev discussion arrived at a "mediated concensus"
of a default feature digit timeout of 1.0 sec. Some voted for 1300;
ctooley thought 1500 for distracted phone users in phone booths;
kpfleming put his foot down at 1.0 sec.
Users who found the previous default max delay of 250 msec perfect,
are welcome to override the new default. Notice that I said that
250 msec was the default; wait a minute, you might say, the config
file said it was 500 msec!; well, because of the bug fix for 14515,
we found that 500 msec was actually enforcing a max of 250. The bug
fix would restore 500 msec, but we felt even that was a bit tight
for most users... 2000 msec was pushed earlier by mmichelson, so
that reduces to 1000 msec after the bug fix. Enjoy!
........
................
This patch prevents the feature detection timeout from being cut in half.
Because the ast_channel_bridge() call will return 0 and pass
a frame pointer for both DTMF_BEGIN and DTMF_END, the feature_timer
field in hte config struct is getting decremented twice, which
effectively cuts the digittimeout in half. I added conditions
to the if statement to only let DTMF_END frames to flow thru,
which solved the problem. Also, when the frame pointer is null,
let control flow thru-- this usually happens on timeouts. I added
a comment to the code to explain what's going on and why.
Many thanks to sodom for reporting this problem. Personnally, it always seemed
like something was wrong with the featuredigittimeout, but I never
could quite decide what... and was too busy to investigate.
This bug forced the issue, and now we know.
Sodom had other issues in 14515, but I couldn't reproduce them. If
he still has problems, and wants to get them solved, he is welcome
to reopen 14515.
Iax2 prune realtime had issues. If "iax2 prune realtime all" was called, it would appear like the command was successful, but in reality nothing happened. This is because the reload that was supposed to take place checks the config files, sees no changes, and does nothing. If there had been a change in the the config file, the realtime users would have been marked for deletion and everything would have been fine. Now prune_users() and prune_peers() are called instead of reload_config() to prune all users/peers that are realtime. These functions remove all users/peers with the rtfriend and delme flags set. iax2_prune_realtime() also lacked the code to properly delete a single friend. For example. if iax2 prune realtime <friend> was called, only the peer instance would be removed. The user would still remain.
Cause astcanary to exit if Asterisk exits abnormally and doesn't kill astcanary.
Also, add some documentation supporting the use of astcanary.
(closes issue #14538)
Reported by: KNK
Patches:
asterisk-1.6.x-astcanary.diff uploaded by KNK (license 545)
........
Allows manager command to see if IAX link is trunked and encrypted. Displays what kind of encryption is enabled as well.
Manager command "iaxpeers" now shows if a link is trunked and encrypted. Instead of encryption saying simply "yes" or "no", it now displays what type of encryption is enabled and if keyrotation is on or not.
Skip check for extension when subscribing for MWI.
Since the remote side is not actually subscribing to a specific extension when
subscribing for MWI just skip the check to see if the extension exists. They can't use it
to specify the mailbox either since we require configuration of that in sip.conf
Russell Bryant [Mon, 23 Feb 2009 23:22:52 +0000 (23:22 +0000)]
Merged revisions 178142 via svnmerge from
https://origsvn.digium.com/svn/asterisk/trunk
................
r178142 | russell | 2009-02-23 17:11:37 -0600 (Mon, 23 Feb 2009) | 22 lines
Merged revisions 178141 via svnmerge from
https://origsvn.digium.com/svn/asterisk/branches/1.4
........
r178141 | russell | 2009-02-23 17:09:01 -0600 (Mon, 23 Feb 2009) | 14 lines
Fix infinite DTMF when a BEGIN is received without an END.
This commit is related to rev 175124 of 1.4 where a previous attempt was made
to fix this problem. The problem with the previous patch was that the inserted
code needed to go _before_ setting the lastrxts to the current timestamp.
Because those were the same, the dtmfcount variable was never decremented, and
so the END was never sent.
In passing, I removed the dtmfsamples variable which was completed unused. I
also removed a redundant setting of the lastrxts variable.
Permit emailsubject and emailbody to be set per mailbox.
(closes issue #14372)
Reported by: fhackenberger
Patches:
voicemail_individual_subject_and_body_1.6.1 uploaded by fhackenberger (license 592)
with additional fixes by Corydon76 (license 14)
........
Key rotation was enabled by default by setting the global encryption method to IAX_ENCRYPT_KEYROTATE. the problem with this is that if encryption is not enabled, and the encryption method is set to anything except 0, the peer appears to have encryption enabled when issuing a "iax2 show peers". Rather than have the key rotation bit always set by default, it is now only set when an encryption method is enabled.
set ASTVARRUNDIR=$(localstatedir)/run/asterisk as default path
When running asterisk as non-root and without this patch the pidfile wants
to go into /var/run/asterisk.pid. This directory is not writable for
the non-root user and changing permissions is not an option.
Putting it in /var/run/asterisk/asterisk.pid makes it possible
to set permissions on the /var/run/asterisk dir so everything
works as it should be.
Prior to spandsp-0.0.6pre4 the t30_stats_t structure used a pages_transferred
integer to indicate the number of pages transferred (so far) during the fax
session. The spandsp-0.0.6pre4 release removed the pages_transferred integer
and replaced it with two different integers - pages_tx and pages_rx. This
revision uses the new integers for spandsp-0.0.6pre4 while maintaining backwards
compatibility for previous spandsp releases.
........
Allow semicolons to be escaped, when passing arguments to the System command.
(closes issue #14231)
Reported by: jcovert
Patches:
20090113__bug14231__2.diff.txt uploaded by Corydon76 (license 14)
corrected_20090113__bug14231__2.diff.txt uploaded by jcovert (license 551)
Tested by: jcovert
........
This exception does not appear to still be true for Solaris 10, and OpenSolaris definitely needs it to be removed.
Fixed for snuff-home on -dev channel.
........
................
David Vossel [Fri, 20 Feb 2009 20:34:11 +0000 (20:34 +0000)]
Fixes issue with undefined audio codecs in chan_iax2
During iax2 call negotiation, supported codecs are passed in an Information Element containing a 2 byte field where each bit correlates to a specific codec. In 1.6 only audio codec bits 0-12 are defined, leaving bits 13-14 undefined. By default all bits are enabled unless specified otherwise. Since its a 2 byte field and 13-14 are not defined, these bits are never turned off. In trunk, bits 13-14 are defined, which means 1.6 is advertising support for codecs it does not have when talking to trunk. I fixed this by adding #define for undefined audio codec bits. These bits are then removed from iax2's full bandwidth capabilities.
This patch fixes a regression of sorts that was introduced in
rev 24425.
It basically fixes AST-190/ABE-1782.
What was wrong: the user has 6000 extensions in one context; and
then 6000 contexts, one per extension. The parser could only handle
about 4893 of the 6000 extens in the single context.
This was due to the regression I mentioned. To get rid of
shift/reduce conflicts, Luigi set up right-recursive lists
for globals, context elements, switch lists, and statements.
Right recursive lists got rid of the warnings, but instead, they
use up a tremendous amount of stack space when the lists are long.
I saw this a few years back, and resolved not to fix it until
someone complained. That day has arrived!
After the changes were made, I ran the regression test suite,
and there were no problems.
I took the test case the user provided, and added 100,000
extensions to the single context, that already had 6,000 extens
in it. (I'll see your 6, and raise you 100!) It takes a few minutes
to read it all in, check it and generate code for it, but no
problems.
So, I think I can say that fundamentally, there are no longer
any limits on the number of items you can place in contexts,
statement blocks, switches, or globals, beyond your virt mem
constraints.
........
................
fix two very minor bugs: if anyone ever uses SLINEAR16 as a format in RTP, ensure that the samples are byte-swapped to network order if needed. also, when a smoother is operating on a format that has a sample rate other than 8000 samples per second, use the proper sample rate for computing delivery timestamps.
........
action_bridge() and bridge_exec() both search for the channels to bridge to, and then immediately drop the lock. Instead, they should hold the lock until the masquerade is complete. This will guarantee the channel remains and prevent any other weirdness from occurring. In action_bridge() some more weirdness comes into play. Both channels are needlessly locked at the same time and perform the exact same logic. It makes sense from a coding organizational standpoint, but could cause a theoretical deadlock so I split the code up. There is an issue associated with this, but since its a rather complicated thing to reproduce I'm not certain this alone will close it.
Modify h323 to build against PTLib as well as the older PWLib
Several changes in PTLib have occurred requiring build time detection. Changes
accounted for include the library name change, config option change, install
location change, and a boolean type change which is handled by ast_ptlib.h.
Also, the sed check has been modified to properly work with autoconf >= 2.62.
This patch fixes merge_contexts_and_delete so it does not deadlock when hints are present.
Reason: when I re-engineered the merge_and_delete func to
reduce its lock time, I failed to notice that the
functions it calls still also do locking as before.
This leads to deadlocks on dialplan reloads, when
there are actually living, subscribed hints registered
in the system.
While the reporter come across this problem while using
AEL, I might note that these deadlocks should also happen
if extensions.conf were used.
All of the above routines are static and restricted
to be used only within pbx.c, and more specifically
within the merge_contexts_and_delete routine.
They are pretty much the same as their counterparts
except they don't lock contexts or hints.
Most of them now do the real work of their
name-alike, with optional locking via extra arguments,
and are called by their name-alike. The goal was to
have the original functions so they would behave
exactly as before.
Both PJ and I tested these fixes, and the deadlocking
problem is no longer encountered.
Russell Bryant [Wed, 18 Feb 2009 06:05:09 +0000 (06:05 +0000)]
Merged revisions 176901 via svnmerge from
https://origsvn.digium.com/svn/asterisk/trunk
........
r176901 | russell | 2009-02-18 00:00:40 -0600 (Wed, 18 Feb 2009) | 9 lines
Fix a number of incorrect uses of strncpy().
The big problem here is that the 3rd argument provided in these uses of strncpy()
did not reserve a byte for the null terminator, leaving the potential for writing
one byte past the end of the buffer.
Aside from this, there were coding guidelines violations with regards to spacing,
as well as hard coded lengths being used instead of sizeof().
T38 faxdetect should jump to the 'fax' extension for incoming calls only
The previous implementation of T38 faxdetect resulted in both sides of the
call jumping to a fax extension when both sides had 't38pt_udptl=yes' and
'faxdetect=yes' in sip.conf and a 'fax' extension in the current context.
This revision will jump to a 'fax' extension on incoming calls only.
........
Several changes to codec_dahdi to play nice with G723.
This commit brings in the changes that were living out on the
svn/asterisk/team/sruffell/asterisk-trunk-transcoder branch. codec_dahdi.c now
always uses signed linear as the simple codec so that a soft g729 codec will
not end up being preferred to the hardware codec. There are also changes to
allow codec_dahdi.c to feed packets to the hardware in the native sample size of
the codec. This solves problems with choppy audio when using G723.
........
create a UDPTL structure in create_addr_from_peer() if it does not already exist for T38
This is required to create a UDPTL structure in create_addr_from_peer() to handle the
scenario where 't38pt_udptl=yes' is not defined in the [general] section of sip.conf but
is defined the peer's context. I tested this patch by enabling t38pt_udptl in the
[general] section on one system and only enabling t38pt_udptl in a peer's context on
the system sending a fax. Without the patch, the sending system will fail to initiate
T38 negotiation with the warning message, "No way to add SDP without an UDPTL structure".
When this patch is applied the sending side will successfully initiate T38 negotiation.
........
Modify bridging to properly evaluate DTMF after first warning is played
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/
........
................
Add assertions in the quest to track down a refcount leak.
(closes issue #14485)
Reported by: davevg
........
r176642 | tilghman | 2009-02-17 15:14:18 -0600 (Tue, 17 Feb 2009) | 8 lines
Prior to masquerade, move the group definitions to the channel performing the
masq, so that the group count lingers past the bridge.
(closes issue #14275)
Reported by: kowalma
Patches:
20090216__bug14275.diff.txt uploaded by Corydon76 (license 14)
Tested by: kowalma
........
Russell Bryant [Tue, 17 Feb 2009 21:05:17 +0000 (21:05 +0000)]
Blocked revisions 176639 via svnmerge
........
r176639 | russell | 2009-02-17 15:04:08 -0600 (Tue, 17 Feb 2009) | 9 lines
Significantly improve scheduler performance under high load.
This patch changes the scheduler to use a max-heap to store pending scheduler
entries instead of a fully sorted doubly linked list. When the number of
entries in the scheduler gets large, this will perform much better. For much
more detailed information on this change, see the review request.
Russell Bryant [Tue, 17 Feb 2009 20:42:04 +0000 (20:42 +0000)]
Blocked revisions 176627 via svnmerge
........
r176627 | russell | 2009-02-17 14:41:24 -0600 (Tue, 17 Feb 2009) | 37 lines
Merge a large set of updates to the Asterisk indications API.
This patch includes a number of changes to the indications API. The primary
motivation for this work was to improve stability. The object management
in this API was significantly flawed, and a number of trivial situations could
cause crashes.
The changes included are:
1) Remove the module res_indications. This included the critical functionality
that actually loaded the indications configuration. I have seen many people
have Asterisk problems because they accidentally did not have an
indications.conf present and loaded. Now, this code is in the core,
and Asterisk will fail to start without indications configuration.
There was one part of res_indications, the dialplan applications, which did
belong in a module, and have been moved to a new module, app_playtones.
2) Object management has been significantly changed. Tone zones are now
managed using astobj2, and it is no longer possible to crash Asterisk by
issuing a reload that destroys tone zones while they are in use.
3) The API documentation has been filled out.
4) The API has been updated to follow our naming conventions.
5) Various bits of code throughout the tree have been updated to account
for the API update.
6) Configuration parsing has been mostly re-written.
7) "Code cleanup"
The code is from svn/asterisk/team/russell/indications/.
Russell Bryant [Tue, 17 Feb 2009 17:38:47 +0000 (17:38 +0000)]
Merged revisions 176557 via svnmerge from
https://origsvn.digium.com/svn/asterisk/trunk
........
r176557 | russell | 2009-02-17 11:33:38 -0600 (Tue, 17 Feb 2009) | 12 lines
Fix a race condition that caused device states to become incorrect for hints.
The problem here is that the hint processing code was subscribed to the wrong
event type. So, it started processing state for a hint too soon, before the
device state cache had been updated.
Also, fix a similar bug in app_queue, as it was also subscribed to the wrong
event type.
After a 'sip reload', qualifies for realtime peers weren't immediately
restarted, instead waiting until the next registration. We're now
caching the qualify across a reload/restart and starting the qualify
immediately upon loading the peer.
(closes issue #14196)
Reported by: pdf
Patches:
20090120__bug14196_1.4.diff.txt uploaded by pdf (license 663)
Tested by: pdf
........
................
Fixes issue with AST_CONTROL_SRCUPDATE not being relayed correctly during bridging
This should have been committed with rev176247, but I missed it. srcupdate frames no longer break out of the native bridge, but are not being sent to the other call leg either. This fixs that.
Use the correct list macros for deleting an item from the middle of a list.
(issue #13777)
Reported by: pj
Patches:
20090203__bug13777.diff.txt uploaded by Corydon76 (license 14)
Tested by: pj
........
projects / thirdparty / asterisk.git / log
