Mark Michelson [Thu, 14 May 2009 22:17:55 +0000 (22:17 +0000)]
Fix a race condition where a reinvite could trigger a 482 response.
The loop detection/spiral detection code in chan_sip used the owner
channel's state as a criterion for determining if the incoming INVITE
is a looped request. The problem with this is that the INVITE-handling
code happens in a different thread than the thread that marks the owner
channel as being up. As a result, if a reinvite were to come in very quickly,
say from another Asterisk on the same LAN, it was possible for the reinvite
to arrive before the owner channel had been set to the up state.
This patch corrects the problem by using the invitestate of the sip_pvt
instead, since that can be guaranteed to be set correctly by the time
the reinvite arrives. Since there is a switch statement further in the
INVITE-handling code, the AST_STATE_RINGING state also checks the invitestate
of the sip_pvt in case we should actually be treating the channel as if it were
up already.
Mark Michelson [Wed, 13 May 2009 19:41:44 +0000 (19:41 +0000)]
Remove an extraneous unlocking operation from ast_channel_free.
In the case that we could not remove the desired channel from the
list of channels, there was an extra call to unlock the channel list.
Since we unlock the list later on in the function anyway, this results
in the list being unlocked twice yet only being locked once.
(closes issue #15098)
Reported by: tim_ringenbach
Patches:
remove_extra_unlock.diff uploaded by tim (license 540)
Tilghman Lesher [Wed, 13 May 2009 00:52:03 +0000 (00:52 +0000)]
Fix logic for how to proceed with a single digit extension.
(closes issue #15091)
Reported by: andrew
Patches:
20090512__issue15091.diff.txt uploaded by tilghman (license 14)
Tested by: andrew
This change modifies app_queue to properly generate CDR records in failure
situations.
This involves setting a proper cdr disposition coresponding to the given
failure condition and ensuring the proper information is stored in the cdr
record.
Mark Michelson [Tue, 12 May 2009 18:18:44 +0000 (18:18 +0000)]
Set the invitestate to INV_CANCELLED only if we are actually sending a SIP CANCEL.
The problem was that the hangup code was setting the invitestate too early. The result of
this was that we would always send a CANCEL request, even if it was not an appropriate
time to do so (e.g. we have not yet received a provisional response for our INVITE).
Note that this same fix had been applied to trunk and the 1.6.X branches starting with
revision 155467. This is why you will see this revision being blocked from those places.
Tilghman Lesher [Mon, 11 May 2009 22:48:20 +0000 (22:48 +0000)]
Move 300 bytes around on the stack, to make more room for an extension buffer.
This allows more concurrent extensions to be copied for a single voicemail,
without creating a possibility of upsetting existing users, where a dialplan
could run out of stack space where it had run fine before. Alternatively,
we could have allocated off the heap, but that is a larger change and would
have increased the chance for instability introduced by this change.
This is really solved starting in 1.6.0.11, as the use of an ast_str buffer
allows an unlimited number of extensions (up to available memory). We
additionally create a new warning message when the buffer length is exceeded,
permitting administrators to see an issue after the fact, whereas previously
the list was silently truncated.
(closes issue #14739)
Reported by: p_lindheimer
Patches:
20090417__bug14739.diff.txt uploaded by tilghman (license 14)
Tested by: p_lindheimer
Richard Mudgett [Mon, 11 May 2009 19:09:00 +0000 (19:09 +0000)]
Sent wrong message to clear a call we started if the other end has not responed yet.
In the state MISDN_CALLING (i.e. SETUP was sent but no answer has arrived yet),
it is not allowed to clear the call with RELEASE_COMPLETE. It must be
cleared with DISCONNECT. A RELEASE_COMPLETE is only allowed as an answer
to a SETUP. (See Q.931 ch. 5.3.2, 5.3.2.a, 5.3.2.b)
Patches:
chan-misdn-ccstate7.patch uploaded by customer.
Make absolute paths for logger channels work properly
(Note: This is not a new feature, it was previously undocumented and broken.)
The Asterisk logger has a feature to support absolute pathnames for logger channels, but the code implementing the feature was broken. This has been fixed, and the absolute path feature is now documented in the sample logger.conf.
Tilghman Lesher [Thu, 7 May 2009 23:41:11 +0000 (23:41 +0000)]
Fix Background within a Macro for FreePBX.
If the single digit DTMF is an extension in the specified context, then
go there and signal no DTMF. Otherwise, we should exit with that DTMF.
If we're in Macro, we'll exit and seek that DTMF as the beginning of an
extension in the Macro's calling context. If we're not in Macro, then
we'll simply seek that extension in the calling context. Previously,
someone complained about the behavior as it related to the interior of a
Gosub routine, and the fix (#14011) inadvertently broke FreePBX
(#14940). This change should fix both of these situations, but with the
possible incompatibility that if a single digit extension does not exist
(but a longer extension COULD have matched), it would have previously
gone immediately to the "i" extension, but will now need to wait for a
timeout.
(closes issue #14940)
Reported by: p_lindheimer
Patches:
20090420__bug14940.diff.txt uploaded by tilghman (license 14)
Tested by: p_lindheimer
Tilghman Lesher [Thu, 7 May 2009 16:29:08 +0000 (16:29 +0000)]
Eliminate repetition of fullcontact during reconstruction.
If the fullcontact field appears in both the sippeers and the
sipregs table, then during reconstruction of the field, it will
otherwise be doubled.
(closes issue #14754)
Reported by: Alexei Gradinari
Patches:
20090506__bug14754.diff.txt uploaded by tilghman (license 14)
Tested by: lmadsen
Jeff Peeler [Wed, 6 May 2009 22:15:19 +0000 (22:15 +0000)]
Make ParkedCall application stop execution of the dialplan after hang up
Just changed park_exec to always return non-zero. I really wasn't entirely sure
at first if this was a bug. Decided it was since it would be surprising when
not using ParkedCall in the dialplan to hang up and have dialplan execution
continue.
Joshua Colp [Tue, 5 May 2009 18:22:27 +0000 (18:22 +0000)]
Fix an incorrect assumption that certain values on the channel will always exist when they may not.
The CDR code involved with bridges wrongly assumed that the currently executing application and data
values will always exist. It is possible for this to be false when call forwarding is involved.
Mark Michelson [Sat, 2 May 2009 10:21:00 +0000 (10:21 +0000)]
Move static buffers to outside for loops in app_chanspy.
Similar to seanbright's commit 191422, this moves some static buffers
to be defined outside of for loops since it is undefined if memory
will be re-used or if the stack will grow with each iteration of the
loop.
Jeff Peeler [Fri, 1 May 2009 17:40:46 +0000 (17:40 +0000)]
Fix DTMF not being sent to other side after a partial feature match
This fixes a regression from commit 176701. The issue was that
ast_generic_bridge never exited after the feature digit timeout had elapsed,
which prevented the queued DTMF from being sent to the other side.
Sean Bright [Fri, 1 May 2009 15:42:48 +0000 (15:42 +0000)]
Move the defintion of the a couple arrays out of loops.
According to Kevin, it is unspecified as to whether a variable defined inside
a block is allocated once by the compiler or for each pass through the block
(loops being the only interesting case), so just define these before we get
into our loop to be sure.
Sean Bright [Wed, 29 Apr 2009 15:23:07 +0000 (15:23 +0000)]
Fix a crash in app_queue with very long member lists.
A user reported via #asterisk that with very long lists of members, a crash
occurs in ast_strdupa, so just use a single buffer and ast_copy_string instead
of stack allocating copys of each interface name.
Kevin P. Fleming [Mon, 27 Apr 2009 19:29:46 +0000 (19:29 +0000)]
Fix 'inconsistent line endings' when autoconf 2.63 is used
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.
Russell Bryant [Mon, 27 Apr 2009 19:00:54 +0000 (19:00 +0000)]
Resolve a crash in res_smdi when used with chan_dahdi.
When chan_dahdi goes to get an SMDI message, it provides no search criteria.
It just grabs the next message that arrives. This code was written with the
SMDI dialplan functions in mind, since that is now the preferred method of
using SMDI. However, this broke support of it being used from chan_dahdi.
If both sides of a Local channel were hung up at around the same time it was
possible for one thread to destroy the local private structure and have the other thread
immediately try to remove the already freed structure from the local channel list.
Detect availability of pthread_rwlock_timedwrlock() before using it.
(closes issue #14930)
Reported by: tilghman
Patches:
20090420__bug14930.diff.txt uploaded by tilghman (license 14)
Tested by: mvanbaak, tilghman
Jeff Peeler [Wed, 22 Apr 2009 19:20:53 +0000 (19:20 +0000)]
Make chan_h323 respect packetization settings
Previously, packetization settings were ignored and now they are not. A new
config option 'autoframing' has been added to mirror the way chan_sip handles
it. Turning on the autoframing option (available both as a global option or per
peer) overrides the local settings with the remote packetization settings.
Testing was performed with varying packetization levels with the following
codecs: ulaw, alaw, gsm, and g729.
(closes issue #12415)
Reported by: pj
Patches:
2009012200_h323packetization.diff.txt uploaded by mvanbaak (license 7),
modified by me
Add check in configure script to check for GLOB_NOMAGIC and GLOB_BRACE in glob.h
This allows config.c to compile when linked against uclibc that does not support these parameters
Add a workaround for func_odbc/ARRAY() for problems that occur with certain special characters.
In certain cases, due to the way Set() works in 1.4, values may not get set
properly. This is a workaround for 1.4 only that corrects for these issues,
without making func_odbc more difficult to use properly.
(closes issue #14614)
Reported by: wdoekes
Patches:
20090309__bug14614__2.diff.txt uploaded by tilghman (license 14)
double_set_unescape_workaround_for_func_odbc.osso-and-tilghman-1.diff uploaded by wdoekes (license 717)
Tested by: wdoekes, tilghman
Sean Bright [Mon, 20 Apr 2009 20:58:39 +0000 (20:58 +0000)]
Properly handle @s within hints in AEL.
AEL was not handling the case of a device hint containing an @ symbol, which
caused parking hints (e.g. hint(park:exten@context)) to error out the parser.
This patch makes AEL treat the @ the same way it treats colon and ampersand
now, meaning the characters are included in verbatim.
Clean up problem with manager implementation of mmap where it was not testing against MAP_FAILED response.
Got rid of shadowed variable used in processign the mmap results.
Change test of mmap results to compare against MAP_FAILED
David Vossel [Sat, 18 Apr 2009 01:27:19 +0000 (01:27 +0000)]
Fixed autologoff in agents.conf not working when agent logs in via AgentLogin app
An agent logs in by calling an extension that calls the AgentLogin app. In agents.conf ackcall=always is set, so when they get a call they have the choice to either acknowledge it or ignore it. autologoff=10 is set as well, so if the agent ignores the call over 10sec one may assume that the agent should be logged out (and in this case hungup on as well), but this was not happening.
Fix a bug where a value used to create the channel name was bogus.
This commit fixes the scenario where an incoming call is authenticated
using a peer entry. Previously the channel name was created using either
the username setting from the sip.conf entry or the IP address that the
call came from. Now the channel name will be created using the peer name
itself. This commit will not change the way the channel name is generated
for users or friends.
(closes issue #14256)
Reported by: Nick_Lewis
Patches:
chan_sip.c-chname.patch uploaded by Nick (license 657)
Tested by: Nick_Lewis, file
David Vossel [Wed, 15 Apr 2009 22:08:40 +0000 (22:08 +0000)]
National prefix inserted even when caller ID not available
When the caller ID is restricted, the expected behavior is for the caller id to be blank. In chan_dahdi, the national prefix is placed onto the callers number even if its restricted (empty) causing the caller id to be the national prefix rather than blank.
David Vossel [Tue, 14 Apr 2009 15:02:04 +0000 (15:02 +0000)]
audio_audiohook_write_list() does not correctly update sample size after ast_translate.
audio_audiohook_write_list() does not take into account that the sample size may change after translation depending on if the original frame is is 8khz or 16khz. While no 16kz codecs are supported in 1.4 at the moment, this will save headaches in the future if they ever are. the sample size is now updated after translating to reflect this possibility. Thanks to jcolp and mmichelson for helping me work this out.
Mark Michelson [Thu, 9 Apr 2009 18:51:20 +0000 (18:51 +0000)]
Handle a SIP race condition (reinvite before an ACK) properly.
RFC 5047 explains the proper course of action to take if a
reINVITE is received before the ACK from a previous invite
transaction. What we are to do is to treat the reINVITE as
if it were both an ACK and a reINVITE and process it normally.
Later, when we receive the ACK we had been expecting, we will
ignore it since its CSeq is less than the current iseqno of
the sip_pvt representing this dialog.
Race condition between ast_cli_command() and 'module unload' could cause a deadlock.
Add lock timeouts to avoid this potential deadlock.
(closes issue #14705)
Reported by: jamessan
Patches:
20090320__bug14705.diff.txt uploaded by tilghman (license 14)
Tested by: jamessan
Backport resolution for file descriptor leak in 1.6.0 to 1.4.
This fixes short reads in http manager sessions, such as those done by the
ast-gui branch. (Fixes AST-198)
Mark Michelson [Wed, 8 Apr 2009 16:52:03 +0000 (16:52 +0000)]
Fix a small logical error when loading moh classes.
We were unconditionally incrementing the number of mohclasses
registered. However, we should actually only increment if the
call to moh_register was successful.
While this probably has never caused problems, I noticed it
and decided to fix it anyway.
Mark Michelson [Wed, 8 Apr 2009 15:26:46 +0000 (15:26 +0000)]
Make a couple of changes with regards to a new message printed in ast_read().
"ast_read() called with no recorded file descriptor" is a new message added
after a bug was discovered. Unfortunately, it seems there are a bunch of places
that potentially make such calls to ast_read() and trigger this error message
to be displayed. This commit does two things to help to make this message appear
less.
First, the message has been downgraded to a debug level message if dev mode is
not enabled. The message means a lot more to developers than it does to end users,
and so developers should take an effort to be sure to call ast_read only when
a channel is ready to be read from. However, since this doesn't actually cause an
error in operation and is not something a user can easily fix, we should not spam
their console with these messages.
Second, the message has been moved to after the check for any pending masquerades.
ast_read() being called with no recorded file descriptor should not interfere with
a masquerade taking place.
This could be seen as a simple way of resolving issue #14723. However, I still want
to try to clear out the existing ways of triggering this message, since I feel that
would be a better resolution for the issue.
Fix a bug where DAHDI/Zaptel channels would not properly switch formats when requested
Don't offer AST_FORMAT_SLINEAR on DAHDI/Zaptel channels... while it could provide a slight performance benefit, the translation core in Asterisk has some flaws when a channel driver offers multiple raw formats. this fix is much simpler than fixing the translation core to solve that issue (although that will be done later).
Distinguish in a sent email between simple sends and forwards.
(closes issue #11678)
Reported by: jamessan
Patches:
20090330__bug11678.diff.txt uploaded by tilghman (license 14)
Tested by: tilghman, lmadsen
Russell Bryant [Fri, 3 Apr 2009 01:57:44 +0000 (01:57 +0000)]
Fix a memory leak in cdr_radius.
I came across this while doing some testing of my ast_channel_ao2 branch.
After running a test overnight that generated over 5 million calls, Asterisk
had taken up about 1 GB of my system memory. So, I re-ran the test with
MALLOC_DEBUG turned on. However, it showed no leaks in Asterisk during the
test, even though Asterisk was still consuming it somehow.
Instead, I turned to valgrind, which when run with --leak-check=full, told
me exactly where the leak came from, which was from allocations inside the
radiusclient-ng library. This explains why MALLOC_DEBUG did not report it.
After a bit of analysis, I found that we were leaking a little bit of memory
every time a CDR record was passed to cdr_radius.
I don't actually have a radius server set up to receive CDR records. However,
I always have my development systems compile and install all modules. In
addition to making sure there are not build errors across modules, always
loading modules helps find bugs like this, too, so it is strongly recommend for
all developers.
the DAHDI_GETCONF, DAHDI_SETCONF and DAHDI_GET_PARAMS ioctls were recently corrected to show that they do, in fact, read data from userspace as part of their work. due to this fix, valgrind now reports a number of cases where chan_dahdi passed an uninitialized (or partially) buffer to these ioctls, which could lead to unexpected behavior.
this patch corrects chan_dahdi to ensure that buffers passed to these ioctls are always fully initialized.
David Vossel [Wed, 1 Apr 2009 19:02:00 +0000 (19:02 +0000)]
Fixes issue with dropped calles due to re-Invite glare and re-Invites never executing after a 491
Acknowledgement for 491 responses were never being processed because it didn't match our pending invite's seqno. Since the ACK was never processed, the 491 frame would continue to be retransmitted until eventually the call was dropped due to max retries. Now during a pending invite, if we receive another invite, we send an 491 and hold on to that glare invite's seqno in the "glareinvite" variable for that sip_pvt struct. When ACK's are received, we first check to see if it is in response to our pending invite, if not we check to see if it is in response to a glare invite. In this case, it is in response to the glare invite and must be dealt with or the call is dropped. I've changed the wait time for resending the re-Invite after receving a 491 response to comply with RFC 3261. Before this patch the scheduled re-Invite would only change a flag indicating that the re-Invite should be sent out, now it actually sends it out as well.
Russell Bryant [Wed, 1 Apr 2009 13:47:30 +0000 (13:47 +0000)]
Fix a case where DTMF could bypass audiohooks.
This change fixes a situation where an audiohook that wants DTMF would not
actually get it. This is in the code path where we end DTMF digit length
emulation while handling a NULL frame.
David Brooks [Tue, 31 Mar 2009 16:37:12 +0000 (16:37 +0000)]
Fix incorrect parsing in chan_gtalk when xmpp contains extra whitespaces
To drill into the xmpp to find the capabilities between channels, chan_gtalk
calls iks_child() and iks_next(). iks_child() and iks_next() are functions in
the iksemel xml parsing library that traverse xml nodes. The bug here is that
both iks_child() and iks_next() will return the next iks_struct node
*regardless* of type. chan_gtalk expects the next node to be of type IKS_TAG,
which in most cases, it is, but in this case (a call being made from the
Empathy IM client), there exists iks_struct nodes which are not IKS_TAG data
(they are extraneous whitespaces), and chan_gtalk doesn't handle that case,
so capabilities don't match, and a call cannot be made.
iks_first_tag() and iks_next_tag(), on the other hand, will not return the
very next iks_struct, but will check to see if the next iks_struct is of
type IKS_TAG. If it isn't, it will be skipped, and the next struct of type
IKS_TAG it finds will be returned. This assures that chan_gtalk will find
the iks_struct it is looking for.
This fix simply changes all calls to iks_child() and iks_next() to become
calls to iks_first_tag() and iks_next_tag(), which resolves the capability
matching.
The following is a payload listing from Empathy, which, due to the extraneous
whitespace, will not be parsed correctly by iksemel:
Mark Michelson [Tue, 31 Mar 2009 15:34:05 +0000 (15:34 +0000)]
Fix some state_interface stuff that was in trunk but not in the backport to 1.4.
Issue #14359 was fixed between the time that I posted the review of the backport
of the state interface change for 1.4. This merges the changes from that issue
back into 1.4.
Richard Mudgett [Mon, 30 Mar 2009 20:38:11 +0000 (20:38 +0000)]
Make chan_misdn BRI TE side normally defer channel selection to the NT side.
Channel allocation collisions are not handled by chan_misdn very well.
This patch simply avoids the problem for BRI only.
For PRI, allocation collisions are still possible but less likely since
there are simply more channels available and each end could use a different
allocation strategy.
misdn.conf options available:
te_choose_channel - Use to force the TE side to allocate channels.
method - Specify the channel allocation strategy.
projects / thirdparty / asterisk.git / log
