It was possible for a reference to a frame which was part of a freed DSP to still be
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.
Add special case for when the agi cannot be executed, to comply with the documentation that
we return failure in that case.
(closes issue #12462)
Reported by: fmueller
Patches:
20080416__bug12462.diff.txt uploaded by Corydon76 (license 14)
Tested by: fmueller
Steve Murphy [Wed, 16 Apr 2008 23:53:27 +0000 (23:53 +0000)]
This is the scariest commit I've done in a long time. This is the astobj2-ification of chan_sip. I've tested a number of scenarios like crazy. It used to have 4x the call setup/teardown performance of trunk, but now it's roughly at parity. I will attempt to find the bottlenecks and get it back to the 4x mark. The changes made were somewhat invasive, but the value to the community of these upgrades outweighs waiting further for more testing. Every change being made to chan_sip was lousing this code up when we tried to merge. Peers, Users, Dialogs, are all now astobj2 objects, indexed via hashtables. Refcounting is used to track objects and free them at the bitter end of their lives. Please file issues on bugs.digium.com, and PLEASE, please, please be patient. One natural advantage to all the hash-table work is that loading large sip.conf files full of thousands of peers now goes much faster. One more please: PLEASE help thrash this code and test it.
Steve Murphy [Wed, 16 Apr 2008 20:54:41 +0000 (20:54 +0000)]
A small enhancement-- I added the routine log_show_lock to utils.c, which if the mentioned lock has been acquired, this routine will log to the console the normal info about that lock you'd see from the CLI when you do a 'core show locks'. It's solely for debug-- if the lock is NOT acquired, there is no output. I use it to show 'unexpected' locks, to see where/why a lock is pre-locked. This command is to be called from points of interest, like just before a trylock, and helps to spot fleeting, highly temporal locks that normally are not locked...
use the ZT_SET_DIALPARAMS ioctl properly by initializing the structure to all zeroes in case it contains fields that we don't write values into (which it does as of Zaptel 1.4.10)
Steve Murphy [Wed, 16 Apr 2008 20:28:08 +0000 (20:28 +0000)]
Introducing a small optimization to event_unsubscribe; events now use a Doubly-Linked list for events, gives fast deletions, for the sake of channel driver mwi events. From team/murf/bug11210.
Steve Murphy [Wed, 16 Apr 2008 20:09:39 +0000 (20:09 +0000)]
Introducing a small upgrade to the ast_sched_xxx facility, to keep it from eating up lots of cpu cycles. See CHANGES. From the team/murf/bug11210 branch.
Steve Murphy [Wed, 16 Apr 2008 17:45:28 +0000 (17:45 +0000)]
Introducing various astobj2 enhancements, chief being a refcount tracing feature, and various documentation updates in astobj2.h, and the addition of standalone utility, refcounter, that will filter the trace output for unbalanced, unfreed objects. This comes from the team/murf/bug11210 branch.
Steve Murphy [Tue, 15 Apr 2008 19:59:50 +0000 (19:59 +0000)]
These changes:
a. fix a self-found problem with SPAWN-ing an extension,
where matches were not being found
b. correct some wording in a comment
c. Add some debug for future debugging.
Allow autofill to work in the general section of queues.conf.
Additionally, don't try to (re)set options when they have empty values in realtime (all unset columns would have an empty value).
Terry Wilson [Mon, 14 Apr 2008 19:12:27 +0000 (19:12 +0000)]
Don't unref user twice on failure. Also, when adding sorted list of users, it is best to check the entry already in the list for a "next" entry instead of the newly created entry...
Increase the retry count when attempting to show channels. This apparently
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
If the datastore has been moved to another channel due to a masquerade, then
freeing the datastore here causes an eventual double free when the new channel
hangs up. We should only free the datastore if we were able to successfully remove
it from the channel we are referencing (i.e. the datastore was not moved).
During hangup it is possible for p->chan or p->owner to be NULL, so just return what the channel is bridged to instead of what they are *really* bridged to. Thanks Matt Nicholson!
Save a local copy of the generate callback prior to unlocking the channel in
case the generate callback goes NULL on us after the channel is unlocked. Thanks
to Russell for pointing this need out to me.
Don't change the SSRC when a new source comes into play, this might happen quite often and depending on the remote side... they might not like this.
(closes issue #12353)
Reported by: dimas
Make sure we start incoming calls on SS7 with echo cancellation enabled. Also make sure when completing a COT we call ss7_start_call with the proper locks held. Lastly, make sure if we fail to get a channel from zt_new that we don't assume it's there.
Use the correct function for free'ing objects, and maybe we won't crash.
(closes issue #12163)
Reported by: gservat
Patches:
20080411__bug12163.diff.txt uploaded by Corydon76 (license 14)
Tested by: gservat
Several places in the code called find_callno() (which releases the lock on the pvt structure) and then immediately locked the call and did things with it. Unfortunately, the call can disappear between the find_callno and the lock, causing Bad Stuff(tm) to happen.
Added find_callno_locked() function to return the callno withtout unlocking for instances that it is needed.
Terry Wilson [Fri, 11 Apr 2008 21:04:46 +0000 (21:04 +0000)]
Fix the fact that global_variables 1) weren't being updated on reload (thanks for the report, Doug), and 2) weren't actually being appended to the list of profile variables because build_profile was called before the list was populated. Also needed to free the contents returned by load_file().
A 'b' option has been added which causes chan_local to return the actual channel that is behind it when queried. This is useful for transfer scenarios as the actual channel will be transferred, not the Local channel. If you have been using Local channels as queue members and having issues when the agent did a blind transfer this option may solve the issue.
Create the directory where name recordings will go if it does not exist.
(closes issue #12311)
Reported by: rkeene
Patches:
12311-mkdir.diff uploaded by qwell (license 4)
Don't hardcode ru into the digits filename so that languageprefix can work.
(closes issue #12404)
Reported by: IgorG
Patches:
voicemail_ru_hardcoded-v1.patch uploaded by IgorG (license 20)
Don't add custom URI options if they don't exist OR they are empty.
(closes issue #12407)
Reported by: homesick
Patches:
uri_options-1.4.diff uploaded by homesick (license 91)
We need to set the persistant_route [sic] parameter for the sip_pvt
during the initial INVITE, no matter if we're building the route set from
an INVITE request or response.
If the [csv] section does not exist in cdr.conf, then an unload/load sequence
is needed to correct the problem. Track whether the load succeeded with a
variable, so we can fix this with a simple reload event, instead.
Mark Michelson [Wed, 9 Apr 2008 17:48:33 +0000 (17:48 +0000)]
There was a subtle logical difference between 1.4 and trunk with regards to how timeouts
were handled. In 1.4, if the absolute timeout were reached on a call, no matter what
the return value of ast_spawn_extension was, the pbx would attempt to go to the 'T'
extension or hangup otherwise. The rearrangement of this function in trunk made this check
only happen in the case that ast_spawn_extension returned 0. If ast_spawn_extension returned
1, then the fact that the timeout expired resulted in a no-op, and would cause an infinite
loop to occur in __ast_pbx_run. This change fixes this problem. Now timeouts will
behave as they did in 1.4
If we receive an AUTHREQ from the remote server and we are unable to reply (for example they have a secret configured, but we do not) then queue a hangup frame on the Asterisk channel. This will cause the channel to hangup and a HANGUP to be sent via IAX2 to the remote side which is the proper thing to do in this scenario.
(closes issue #12385)
Reported by: viraptor
Permit callee to continue in the dialplan, after caller has hung up.
(closes issue #11954)
Reported by: johan
Patches:
app_dial_rev104031.patch uploaded by johan (license 334)
Add commandline tool for doing CLI commands through AMI (instead of using asterisk -rx)
(closes issue #12389)
Reported by: davevg
Patches:
astcli uploaded by davevg (license 209)
Fix potential buffer overflow that could happen if more than 100 announce files
were specified when calling ParkAndAnnounce. This overflow is not exploitable remotely
and so there is no need for a security advisory.
Move check for still-bridged channels out a little further, to avoid possible
deadlocks. (Closes issue #12252)
Reported by: callguy
Patches:
20080319__bug12252.diff.txt uploaded by Corydon76 (license 14)
Tested by: callguy
If audio suddenly gets fed into one side of a channel after a lapse of frames flush the other factory so that old audio does not remain in the factory causing the sync code to not execute.
(closes issue #12296)
Reported by: jvandal
This one line change makes an if inside a for loop (in realtime_peer) check all the ast_variables the loop was intending to test rather than just the first one.
Mark Michelson [Mon, 7 Apr 2008 20:22:06 +0000 (20:22 +0000)]
This is a "fix" for something that's been bugging the crap out of me for a while.
The variable name "flag" to distinguish between whether a message is being forwarded or
is new is not a helpful name. The newly added doxygen documentation to app_voicemail is
tremendously helpful, but I still just...hate this variable name. I think is_new_message
is more indicative of what its purpose is.
This fix prevents a deadlock that was experienced in chan_local. There was
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.
This one line change makes an if inside a for loop (in realtime_peer) check all the ast_variables the loop was intending to test rather than just the first one.
AsyncAGI should not close the manager session on error.
(closes issue #12370)
Reported by: srt
Patches:
asterisk-12370.diff uploaded by srt (license 378)
projects / thirdparty / asterisk.git / log
