Two fixes:
1) One must always have the private unlocked before calling
pbx_builtin_setvar_helper to not invalidate locking order since it locks the
channel.
2) Unlock the channel before calling pbx_find_extension, which starts and stops
autoservice during the lookup. The problem scenario as illustrated by the
reporter:
Thread: do_monitor
-----------------------
handle_request_do
handle_incoming
handle_request_refer
ast_parking_ext_valid
pbx_find_extension
ast_autoservice_stop
while (chan_list_state == as_chan_list_state) { usleep(1000); }
Thread: autoservice_run
-----------------------
autoservice_run
chan = ast_waitfor_n
ast_waitfor_nandfds
ast_waitfor_nandfds_classic / simple / complex (depending on your system)
ast_channel_lock(c[x]);
handle_request_do and schedule_process_request_queue locks the owner
if it exists. The autoservice thread is waiting for the channel lock, which
wasn't ever released since the do_monitor thread was waiting for autoservice
operations to complete. Solved by unlocking the channel but keeping a reference
to guarantee safety.
(closes issue #18403)
Reported by: jthurman
Patches:
20110103-blind_deadlock.diff uploaded by jthurman (license 614)
issue18403.patch uploaded by jpeeler (license 325)
Tested by: jthurman
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.8@301790 65c4cc65-6c06-0410-ace0-fbb531ad65f3
https://origsvn.digium.com/svn/asterisk/branches/1.6.2
................
r301594 | mnicholson | 2011-01-12 12:50:31 -0600 (Wed, 12 Jan 2011) | 15 lines
Removed a usleep(1) that shouldn't be necessary in session_do, and removed the
ms_t member from the mansession_session structure.
Merged revisions 301591 via svnmerge from
https://origsvn.digium.com/svn/asterisk/branches/1.4
........
r301591 | mnicholson | 2011-01-12 12:39:03 -0600 (Wed, 12 Jan 2011) | 5 lines
Don't store the thread id for the manager session in the structure we pass to
the thread for the manager session.
ABE-2543
........
................
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.8@301595 65c4cc65-6c06-0410-ace0-fbb531ad65f3
https://origsvn.digium.com/svn/asterisk/branches/1.6.2
................
r301503 | jpeeler | 2011-01-12 12:11:49 -0600 (Wed, 12 Jan 2011) | 19 lines
Merged revisions 301502 via svnmerge from
https://origsvn.digium.com/svn/asterisk/branches/1.4
........
r301502 | jpeeler | 2011-01-12 12:10:42 -0600 (Wed, 12 Jan 2011) | 12 lines
Fix CPU spike when pressing DTMF after agent login.
The problem here is that DTMF was being continuously deferred and requeued
since ast_safe_sleep is called in a loop. There are serveral other places in the
code that sleeps and then loops in a similar fashion. Because of this fact I
opted to not defer DTMF any more, which will not affect the original fix:
https://reviewboard.asterisk.org/r/674
(closes issue #18130)
Reported by: rgj
........
................
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.8@301504 65c4cc65-6c06-0410-ace0-fbb531ad65f3
the warning, if no user interface for menuselect warning was found is not right.
you have to rerun configure before make menuselect after installing a proper user interface.
(closes issue #18594)
Reported by: Dovid
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.8@301444 65c4cc65-6c06-0410-ace0-fbb531ad65f3
https://origsvn.digium.com/svn/asterisk/branches/1.6.2
........
r301220 | pabelanger | 2011-01-09 16:38:24 -0500 (Sun, 09 Jan 2011) | 14 lines
SOUND_CACHE_DIR now defaults to empty
Sounds files included in the Asterisk tarball were being ignored and
re-downloaded. Users wanting to cache the files can still override the setting
using the --with-sounds-cache option.
(closes issue #18589)
Reported by: pabelanger
Patches:
issue18589.patch uploaded by pabelanger (license 224)
Tested by: pabelanger
Review: https://reviewboard.asterisk.org/r/1074/
........
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.8@301221 65c4cc65-6c06-0410-ace0-fbb531ad65f3
The DAHDI ISDN channel name is not dialable.
Make a channel name like DAHDI/i3/400-12 dialable when the sequence number
is stripped off of the name.
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.8@301134 65c4cc65-6c06-0410-ace0-fbb531ad65f3
https://origsvn.digium.com/svn/asterisk/branches/1.6.2
........
r301046 | jpeeler | 2011-01-07 13:57:42 -0600 (Fri, 07 Jan 2011) | 8 lines
Fix regression causing forwarding voicemails to not work with file storage.
I had actually already fixed this in 295200 in 1.4 and thought it wasn't
missing in the other branches for some reason.
(closes issue #18358)
Reported by: cabal95
........
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.8@301047 65c4cc65-6c06-0410-ace0-fbb531ad65f3
https://origsvn.digium.com/svn/asterisk/branches/1.6.2
................
r300951 | jpeeler | 2011-01-07 11:23:37 -0600 (Fri, 07 Jan 2011) | 14 lines
Merged revisions 300918 via svnmerge from
https://origsvn.digium.com/svn/asterisk/branches/1.4
........
r300918 | jpeeler | 2011-01-07 11:13:21 -0600 (Fri, 07 Jan 2011) | 7 lines
Ensure good bye prompt in voicemail is played at the correct time.
Specifically in the case of timing out but not leaving voicemail nothing
should be heard. And when leaving voicemail it should be heard.
ABE-2647
........
................
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.8@300955 65c4cc65-6c06-0410-ace0-fbb531ad65f3
https://origsvn.digium.com/svn/asterisk/be/branches/C.3-bier
..........
r300711 | rmudgett | 2011-01-05 13:43:55 -0600 (Wed, 05 Jan 2011) | 14 lines
A call retrieved from hold may wind up with no audio.
If the retrieved call is natively bridged then the call may not have any
audio path. The following warning message is given:
"Failed to add <dfd> to conference <chan>/<chan>: Invalid argument".
* Open the media on a B channel when pri_fixup_principle() moves the call
from a no_b_channel channel to a real channel.
* Added lock protection while pri_fixup_principle() moves a call from one
private structure to another.
* Made some pri_fixup_principle() messages more meaningful.
..........
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.8@300714 65c4cc65-6c06-0410-ace0-fbb531ad65f3
https://origsvn.digium.com/svn/asterisk/branches/1.6.2
................
r300622 | tilghman | 2011-01-05 12:54:58 -0600 (Wed, 05 Jan 2011) | 17 lines
Merged revisions 300621 via svnmerge from
https://origsvn.digium.com/svn/asterisk/branches/1.4
........
r300621 | tilghman | 2011-01-05 12:47:46 -0600 (Wed, 05 Jan 2011) | 10 lines
Use the sanity check in place of the disconnect/connect cycle.
The disconnect/connect cycle has the potential to cause random crashes.
(closes issue #18243)
Reported by: ks3
Patches:
res_odbc.patch uploaded by ks3 (license 1147)
Tested by: ks3
........
................
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.8@300623 65c4cc65-6c06-0410-ace0-fbb531ad65f3
https://origsvn.digium.com/svn/asterisk/branches/1.6.2
................
r300298 | twilson | 2011-01-04 11:37:26 -0600 (Tue, 04 Jan 2011) | 22 lines
Merged revisions 300216 via svnmerge from
https://origsvn.digium.com/svn/asterisk/branches/1.4
........
r300216 | twilson | 2011-01-04 11:11:48 -0600 (Tue, 04 Jan 2011) | 15 lines
Don't authenticate SUBSCRIBE re-transmissions
This only skips authentication on retransmissions that are already
authenticated. A similar method is already used for INVITES. This
is the kind of thing we end up having to do when we don't have a
transaction layer...
(closes issue #18075)
Reported by: mdu113
Patches:
diff.txt uploaded by twilson (license 396)
Tested by: twilson, mdu113
Review: https://reviewboard.asterisk.org/r/1005/
........
................
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.8@300301 65c4cc65-6c06-0410-ace0-fbb531ad65f3
I've increased the size of the response field in a DUNDi mapping because of
some documentation I'm writing. Previously it was set to AST_MAX_EXTENSION which
is only 80 characters, which is far too small when you're using some dialplan
functions to craft a response. The example I'm using is:
extensions =>
RegisteredDevices,0,SIP,dundi:very_awesome_password/${IF($[${DB_EXISTS(phones/${NUMBER}/device)}]?${DB(phones/${NUMBER}/device)}:None)},nopartial
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.8@300082 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Change order of sending Terminal Capability Set and MasterSlave
Determination packets, MSD send when TCS exchange procedure is done
(we send tcs ack to remote and we have remote tcs ack already
or we receive tcs ack from remote and we have send our tcs ack to
remote already). Some endpoints can work in this sequence only,
i suggest they can't work with both (tcs and msd) exchange procedures
simultaneously.
Also changed StartH245 facility message sending. It send on
incoming calls only due to some endpoints can't proccess properly
this facility messages on their incoming calls.
(issue #18433)
Reported by: MrHanMan
Patches:
tcs-msd-h245-3.patch uploaded by may213 (license 454)
Tested by: MrHanMan, may213
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.8@299711 65c4cc65-6c06-0410-ace0-fbb531ad65f3
........
r299533 | moy | 2010-12-22 22:02:31 -0500 (Wed, 22 Dec 2010) | 1 line
do not use progress which is for PRI and SS7, add mfcr2_progress member
........
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.8@299573 65c4cc65-6c06-0410-ace0-fbb531ad65f3
https://origsvn.digium.com/svn/asterisk/branches/1.6.2
........
r299448 | tilghman | 2010-12-22 14:03:30 -0600 (Wed, 22 Dec 2010) | 8 lines
Resolve warnings by disambiguating the "s" extension as used by chan_dahdi from the "s" extension as used by the AEL macros.
(closes issue #18480)
Reported by: nivek
Patches:
20101215__issue18480__2.diff.txt uploaded by tilghman (license 14)
Tested by: nivek
........
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.8@299449 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Chan_dahdi always inserts a connected party IE when you call from one
dahdi channel to another dahdi channel, even if no such information was
received on the 2nd channel. This clears the display of many phones.
* Removed leftover artifact from before the valid flag was added.
* Updated all of the channel's caller id information with the new
connected line information instead of just the string parts.
(closes issue #18508)
Reported by: wimpy
Patches:
issue18508_trunk.patch uploaded by rmudgett (license 664)
Tested by: wimpy, rmudgett
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.8@299405 65c4cc65-6c06-0410-ace0-fbb531ad65f3
https://origsvn.digium.com/svn/asterisk/branches/1.6.2
................
r299242 | mnicholson | 2010-12-20 15:25:35 -0600 (Mon, 20 Dec 2010) | 23 lines
Merged revisions 299194,299198,299220 via svnmerge from
https://origsvn.digium.com/svn/asterisk/branches/1.4
........
r299194 | mnicholson | 2010-12-20 14:45:38 -0600 (Mon, 20 Dec 2010) | 6 lines
Respond as soon as possible with a 202 Accepted to refer requests.
This change also plugs a few memory leaks that can occur when parking sip calls.
ABE-2656
........
r299198 | mnicholson | 2010-12-20 15:00:44 -0600 (Mon, 20 Dec 2010) | 2 lines
Remove changes to via processing that were not supposed to go into the last commit.
........
r299220 | mnicholson | 2010-12-20 15:21:39 -0600 (Mon, 20 Dec 2010) | 4 lines
Use ast_free() instead of free()
ABE-2656
........
................
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.8@299353 65c4cc65-6c06-0410-ace0-fbb531ad65f3
* Make sure to allocate a cc_params structure
when creating autopeers.
* Use sip_uri_cmp when retrieving SIP CC agents
and monitors in case parameters appear in the
URI.
(closes issue #18504)
Reported by: kkm
(closes issue #18338)
Reported by: GeorgeKonopacki
Patches:
18338.diff uploaded by mmichelson (license 60)
Tested by: GeorgeKonopacki
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.8@299248 65c4cc65-6c06-0410-ace0-fbb531ad65f3
https://origsvn.digium.com/svn/asterisk/branches/1.6.2
........
r299130 | tilghman | 2010-12-20 11:41:24 -0600 (Mon, 20 Dec 2010) | 11 lines
If a call was not answered, then the billsec was calculated unusually large.
Also, due to a copy and paste error, a request for the answer field would have
given the start value, instead.
(closes issue #18460)
Reported by: joscas
Patches:
20101215__issue18460.diff.txt uploaded by tilghman (license 14)
Tested by: joscas
........
git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.8@299131 65c4cc65-6c06-0410-ace0-fbb531ad65f3
