Age | Commit message (Collapse) | Author |
|
Based on feedback from Corey Farrell and Y Ateya, a few new
macros have been added...
AST_VECTOR_REMOVE which takes a parameter to indicate if
order should be preserved.
AST_VECTOR_ADD_SORTED which adds an element to
a sorted vector.
AST_VECTOR_RESET which cleans all elements from the vector
leaving the storage intact.
Change-Id: I41d32dbdf7137e0557134efeff9f9f1064b58d14
|
|
After using the new vector stuff for real I found...
A bug in AST_VECTOR_INSERT_AT that could cause a seg fault.
The callbacks needed to be closer to ao2_callback in behavior
WRT to CMP_MATCH and CMP_STOP behavior and the ability to return
a vector of matched entries.
A pre-existing issue with APPEND and REPLACE was also fixed.
I also added a new macro to test.h that acts like ast_test_validate
but also accepts a return code variable and a cleanup label. As well
as printing the error, it sets the rc variable to AST_TEST_FAIL and
does a goto to the specified label on error. I had a local version
of this in test_vector so I just moved it.
ASTERISK-25045
Change-Id: I05e5e47fd02f61964be13b7e8942bab5d61b29cc
|
|
Renamed AST_VECTOR_INSERT to AST_VECTOR_REPLACE because it really
does replace not insert. The few users of AST_VECTOR_INSERT were
refactored. Because these are macros, there should be no ABI
compatibility issues.
Added AST_VECTOR_INSERT_AT that actually inserts an element into the
vector at a specific index pushing existing elements to the right.
Added AST_VECTOR_GET_CMP that can retrieve from the vector based
on a user-provided compare function.
Added AST_VECTOR_CALLBACK function that will execute a function
for each element in the vector. Similar to ao2_callback and
ao2_callback_data functions although the vector callback can take
a variable number of arguments. This should allow easy migration
to a vector where a container might be too heavy.
Added read/write locked vector and lock manipulation macros.
Added unit tests.
ASTERISK-25045 #close
Change-Id: I2e07ecc709d2f5f91bcab8904e5e9340609b00e0
|
|
In the old times media formats were represented using a bit field. This was
fast but had a few limitations.
1. Asterisk was limited in how many formats it could handle.
2. Formats, being a bit field, could not include any attribute information.
A format was strictly its type, e.g., "this is ulaw".
This was changed in Asterisk 10 (see
https://wiki.asterisk.org/wiki/display/AST/Media+Architecture+Proposal for
notes on that work) which led to the creation of the ast_format structure.
This structure allowed Asterisk to handle attributes and bundle information
with a format.
Additionally, ast_format_cap was created to act as a container for multiple
formats that, together, formed the capability of some entity. Another
mechanism was added to allow logic to be registered which performed format
attribute negotiation. Everywhere throughout the codebase Asterisk was
changed to use this strategy.
Unfortunately, in software, there is no free lunch. These new capabilities
came at a cost.
Performance analysis and profiling showed that we spend an inordinate
amount of time comparing, copying, and generally manipulating formats and
their related structures. Basic prototyping has shown that a reasonably
large performance improvement could be made in this area. This patch is the
result of that project, which overhauled the media format architecture
and its usage in Asterisk to improve performance.
Generally, the new philosophy for handling formats is as follows:
* The ast_format structure is reference counted. This removed a large amount
of the memory allocations and copying that was done in prior versions.
* In order to prevent race conditions while keeping things performant, the
ast_format structure is immutable by convention and lock-free. Violate this
tenet at your peril!
* Because formats are reference counted, codecs are also reference counted.
The Asterisk core generally provides built-in codecs and caches the
ast_format structures created to represent them. Generally, to prevent
inordinate amounts of module reference bumping, codecs and formats can be
added at run-time but cannot be removed.
* All compatibility with the bit field representation of codecs/formats has
been moved to a compatibility API. The primary user of this representation
is chan_iax2, which must continue to maintain its bit-field usage of formats
for interoperability concerns.
* When a format is negotiated with attributes, or when a format cannot be
represented by one of the cached formats, a new format object is created or
cloned from an existing format. That format may have the same codec
underlying it, but is a different format than a version of the format with
different attributes or without attributes.
* While formats are reference counted objects, the reference count maintained
on the format should be manipulated with care. Formats are generally cached
and will persist for the lifetime of Asterisk and do not explicitly need
to have their lifetime modified. An exception to this is when the user of a
format does not know where the format came from *and* the user may outlive
the provider of the format. This occurs, for example, when a format is read
from a channel: the channel may have a format with attributes (hence,
non-cached) and the user of the format may last longer than the channel (if
the reference to the channel is released prior to the format's reference).
For more information on this work, see the API design notes:
https://wiki.asterisk.org/wiki/display/AST/Media+Format+Rewrite
Finally, this work was the culmination of a large number of developer's
efforts. Extra thanks goes to Corey Farrell, who took on a large amount of the
work in the Asterisk core, chan_sip, and was an invaluable resource in peer
reviews throughout this project.
There were a substantial number of patches contributed during this work; the
following issues/patch names simply reflect some of the work (and will cause
the release scripts to give attribution to the individuals who work on them).
Reviews:
https://reviewboard.asterisk.org/r/3814
https://reviewboard.asterisk.org/r/3808
https://reviewboard.asterisk.org/r/3805
https://reviewboard.asterisk.org/r/3803
https://reviewboard.asterisk.org/r/3801
https://reviewboard.asterisk.org/r/3798
https://reviewboard.asterisk.org/r/3800
https://reviewboard.asterisk.org/r/3794
https://reviewboard.asterisk.org/r/3793
https://reviewboard.asterisk.org/r/3792
https://reviewboard.asterisk.org/r/3791
https://reviewboard.asterisk.org/r/3790
https://reviewboard.asterisk.org/r/3789
https://reviewboard.asterisk.org/r/3788
https://reviewboard.asterisk.org/r/3787
https://reviewboard.asterisk.org/r/3786
https://reviewboard.asterisk.org/r/3784
https://reviewboard.asterisk.org/r/3783
https://reviewboard.asterisk.org/r/3778
https://reviewboard.asterisk.org/r/3774
https://reviewboard.asterisk.org/r/3775
https://reviewboard.asterisk.org/r/3772
https://reviewboard.asterisk.org/r/3761
https://reviewboard.asterisk.org/r/3754
https://reviewboard.asterisk.org/r/3753
https://reviewboard.asterisk.org/r/3751
https://reviewboard.asterisk.org/r/3750
https://reviewboard.asterisk.org/r/3748
https://reviewboard.asterisk.org/r/3747
https://reviewboard.asterisk.org/r/3746
https://reviewboard.asterisk.org/r/3742
https://reviewboard.asterisk.org/r/3740
https://reviewboard.asterisk.org/r/3739
https://reviewboard.asterisk.org/r/3738
https://reviewboard.asterisk.org/r/3737
https://reviewboard.asterisk.org/r/3736
https://reviewboard.asterisk.org/r/3734
https://reviewboard.asterisk.org/r/3722
https://reviewboard.asterisk.org/r/3713
https://reviewboard.asterisk.org/r/3703
https://reviewboard.asterisk.org/r/3689
https://reviewboard.asterisk.org/r/3687
https://reviewboard.asterisk.org/r/3674
https://reviewboard.asterisk.org/r/3671
https://reviewboard.asterisk.org/r/3667
https://reviewboard.asterisk.org/r/3665
https://reviewboard.asterisk.org/r/3625
https://reviewboard.asterisk.org/r/3602
https://reviewboard.asterisk.org/r/3519
https://reviewboard.asterisk.org/r/3518
https://reviewboard.asterisk.org/r/3516
https://reviewboard.asterisk.org/r/3515
https://reviewboard.asterisk.org/r/3512
https://reviewboard.asterisk.org/r/3506
https://reviewboard.asterisk.org/r/3413
https://reviewboard.asterisk.org/r/3410
https://reviewboard.asterisk.org/r/3387
https://reviewboard.asterisk.org/r/3388
https://reviewboard.asterisk.org/r/3389
https://reviewboard.asterisk.org/r/3390
https://reviewboard.asterisk.org/r/3321
https://reviewboard.asterisk.org/r/3320
https://reviewboard.asterisk.org/r/3319
https://reviewboard.asterisk.org/r/3318
https://reviewboard.asterisk.org/r/3266
https://reviewboard.asterisk.org/r/3265
https://reviewboard.asterisk.org/r/3234
https://reviewboard.asterisk.org/r/3178
ASTERISK-23114 #close
Reported by: mjordan
media_formats_translation_core.diff uploaded by kharwell (License 6464)
rb3506.diff uploaded by mjordan (License 6283)
media_format_app_file.diff uploaded by kharwell (License 6464)
misc-2.diff uploaded by file (License 5000)
chan_mild-3.diff uploaded by file (License 5000)
chan_obscure.diff uploaded by file (License 5000)
jingle.diff uploaded by file (License 5000)
funcs.diff uploaded by file (License 5000)
formats.diff uploaded by file (License 5000)
core.diff uploaded by file (License 5000)
bridges.diff uploaded by file (License 5000)
mf-codecs-2.diff uploaded by file (License 5000)
mf-app_fax.diff uploaded by file (License 5000)
mf-apps-3.diff uploaded by file (License 5000)
media-formats-3.diff uploaded by file (License 5000)
ASTERISK-23715
rb3713.patch uploaded by coreyfarrell (License 5909)
rb3689.patch uploaded by mjordan (License 6283)
ASTERISK-23957
rb3722.patch uploaded by mjordan (License 6283)
mf-attributes-3.diff uploaded by file (License 5000)
ASTERISK-23958
Tested by: jrose
rb3822.patch uploaded by coreyfarrell (License 5909)
rb3800.patch uploaded by jrose (License 6182)
chan_sip.diff uploaded by mjordan (License 6283)
rb3747.patch uploaded by jrose (License 6182)
ASTERISK-23959 #close
Tested by: sgriepentrog, mjordan, coreyfarrell
sip_cleanup.diff uploaded by opticron (License 6273)
chan_sip_caps.diff uploaded by mjordan (License 6283)
rb3751.patch uploaded by coreyfarrell (License 5909)
chan_sip-3.diff uploaded by file (License 5000)
ASTERISK-23960 #close
Tested by: opticron
direct_media.diff uploaded by opticron (License 6273)
pjsip-direct-media.diff uploaded by file (License 5000)
format_cap_remove.diff uploaded by opticron (License 6273)
media_format_fixes.diff uploaded by opticron (License 6273)
chan_pjsip-2.diff uploaded by file (License 5000)
ASTERISK-23966 #close
Tested by: rmudgett
rb3803.patch uploaded by rmudgetti (License 5621)
chan_dahdi.diff uploaded by file (License 5000)
ASTERISK-24064 #close
Tested by: coreyfarrell, mjordan, opticron, file, rmudgett, sgriepentrog, jrose
rb3814.patch uploaded by rmudgett (License 5621)
moh_cleanup.diff uploaded by opticron (License 6273)
bridge_leak.diff uploaded by opticron (License 6273)
translate.diff uploaded by file (License 5000)
rb3795.patch uploaded by rmudgett (License 5621)
tls_fix.diff uploaded by mjordan (License 6283)
fax-mf-fix-2.diff uploaded by file (License 5000)
rtp_transfer_stuff uploaded by mjordan (License 6283)
rb3787.patch uploaded by rmudgett (License 5621)
media-formats-explicit-translate-format-3.diff uploaded by file (License 5000)
format_cache_case_fix.diff uploaded by opticron (License 6273)
rb3774.patch uploaded by rmudgett (License 5621)
rb3775.patch uploaded by rmudgett (License 5621)
rtp_engine_fix.diff uploaded by opticron (License 6273)
rtp_crash_fix.diff uploaded by opticron (License 6273)
rb3753.patch uploaded by mjordan (License 6283)
rb3750.patch uploaded by mjordan (License 6283)
rb3748.patch uploaded by rmudgett (License 5621)
media_format_fixes.diff uploaded by opticron (License 6273)
rb3740.patch uploaded by mjordan (License 6283)
rb3739.patch uploaded by mjordan (License 6283)
rb3734.patch uploaded by mjordan (License 6283)
rb3689.patch uploaded by mjordan (License 6283)
rb3674.patch uploaded by coreyfarrell (License 5909)
rb3671.patch uploaded by coreyfarrell (License 5909)
rb3667.patch uploaded by coreyfarrell (License 5909)
rb3665.patch uploaded by mjordan (License 6283)
rb3625.patch uploaded by coreyfarrell (License 5909)
rb3602.patch uploaded by coreyfarrell (License 5909)
format_compatibility-2.diff uploaded by file (License 5000)
core.diff uploaded by file (License 5000)
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@419044 65c4cc65-6c06-0410-ace0-fbb531ad65f3
|
|
C does not support templates like C++.
........
Merged revisions 402438 from http://svn.asterisk.org/svn/asterisk/branches/12
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@402439 65c4cc65-6c06-0410-ace0-fbb531ad65f3
|
|
Made the vector macro API be more like linked lists.
1) Added a name parameter to ast_vector() to name the vector struct.
2) Made the API take a pointer to the vector struct instead of the struct
itself.
3) Added an element cleanup macro/function parameter when removing an
element from the vector for ast_vector_remove_cmp_unordered() and
ast_vector_remove_elem_unordered().
4) Added ast_vector_get_addr() in case the vector element is not a simple
pointer.
* Converted an inline vector usage in stasis_message_router to use the
vector API. It needed the API improvements so it could be converted.
* Fixed topic reference leak in router_dtor() when the
stasis_message_router is destroyed.
* Fixed deadlock potential in stasis_forward_all() and
stasis_forward_cancel(). Locking two topics at the same time requires
deadlock avoidance.
* Made internal_stasis_subscribe() tolerant of a NULL topic.
* Made stasis_message_router_add(),
stasis_message_router_add_cache_update(), stasis_message_router_remove(),
and stasis_message_router_remove_cache_update() tolerant of a NULL
message_type.
* Promoted a LOG_DEBUG message to LOG_ERROR as intended in
dispatch_message().
Review: https://reviewboard.asterisk.org/r/2903/
........
Merged revisions 402429 from http://svn.asterisk.org/svn/asterisk/branches/12
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@402430 65c4cc65-6c06-0410-ace0-fbb531ad65f3
|
|
........
r399887 | dlee | 2013-09-26 10:41:47 -0500 (Thu, 26 Sep 2013) | 1 line
Minor performance bump by not allocate manager variable struct if we don't need it
........
r400138 | dlee | 2013-09-30 10:24:00 -0500 (Mon, 30 Sep 2013) | 23 lines
Stasis performance improvements
This patch addresses several performance problems that were found in
the initial performance testing of Asterisk 12.
The Stasis dispatch object was allocated as an AO2 object, even though
it has a very confined lifecycle. This was replaced with a straight
ast_malloc().
The Stasis message router was spending an inordinate amount of time
searching hash tables. In this case, most of our routers had 6 or
fewer routes in them to begin with. This was replaced with an array
that's searched linearly for the route.
We more heavily rely on AO2 objects in Asterisk 12, and the memset()
in ao2_ref() actually became noticeable on the profile. This was
#ifdef'ed to only run when AO2_DEBUG was enabled.
After being misled by an erroneous comment in taskprocessor.c during
profiling, the wrong comment was removed.
Review: https://reviewboard.asterisk.org/r/2873/
........
r400178 | dlee | 2013-09-30 13:26:27 -0500 (Mon, 30 Sep 2013) | 24 lines
Taskprocessor optimization; switch Stasis to use taskprocessors
This patch optimizes taskprocessor to use a semaphore for signaling,
which the OS can do a better job at managing contention and waiting
that we can with a mutex and condition.
The taskprocessor execution was also slightly optimized to reduce the
number of locks taken.
The only observable difference in the taskprocessor implementation is
that when the final reference to the taskprocessor goes away, it will
execute all tasks to completion instead of discarding the unexecuted
tasks.
For systems where unnamed semaphores are not supported, a really
simple semaphore implementation is provided. (Which gives identical
performance as the original taskprocessor implementation).
The way we ended up implementing Stasis caused the threadpool to be a
burden instead of a boost to performance. This was switched to just
use taskprocessors directly for subscriptions.
Review: https://reviewboard.asterisk.org/r/2881/
........
r400180 | dlee | 2013-09-30 13:39:34 -0500 (Mon, 30 Sep 2013) | 28 lines
Optimize how Stasis forwards are dispatched
This patch optimizes how forwards are dispatched in Stasis.
Originally, forwards were dispatched as subscriptions that are invoked
on the publishing thread. This did not account for the vast number of
forwards we would end up having in the system, and the amount of work it
would take to walk though the forward subscriptions.
This patch modifies Stasis so that rather than walking the tree of
forwards on every dispatch, when forwards and subscriptions are changed,
the subscriber list for every topic in the tree is changed.
This has a couple of benefits. First, this reduces the workload of
dispatching messages. It also reduces contention when dispatching to
different topics that happen to forward to the same aggregation topic
(as happens with all of the channel, bridge and endpoint topics).
Since forwards are no longer subscriptions, the bulk of this patch is
simply changing stasis_subscription objects to stasis_forward objects
(which, admittedly, I should have done in the first place.)
Since this required me to yet again put in a growing array, I finally
abstracted that out into a set of ast_vector macros in
asterisk/vector.h.
Review: https://reviewboard.asterisk.org/r/2883/
........
r400181 | dlee | 2013-09-30 13:48:57 -0500 (Mon, 30 Sep 2013) | 28 lines
Remove dispatch object allocation from Stasis publishing
While looking for areas for performance improvement, I realized that an
unused feature in Stasis was negatively impacting performance.
When a message is sent to a subscriber, a dispatch object is allocated
for the dispatch, containing the topic the message was published to, the
subscriber the message is being sent to, and the message itself.
The topic is actually unused by any subscriber in Asterisk today. And
the subscriber is associated with the taskprocessor the message is being
dispatched to.
First, this patch removes the unused topic parameter from Stasis
subscription callbacks.
Second, this patch introduces the concept of taskprocessor local data,
data that may be set on a taskprocessor and provided along with the data
pointer when a task is pushed using the ast_taskprocessor_push_local()
call. This allows the task to have both data specific to that
taskprocessor, in addition to data specific to that invocation.
With those two changes, the dispatch object can be removed completely,
and the message is simply refcounted and sent directly to the
taskprocessor.
Review: https://reviewboard.asterisk.org/r/2884/
........
Merged revisions 399887,400138,400178,400180-400181 from http://svn.asterisk.org/svn/asterisk/branches/12
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@400186 65c4cc65-6c06-0410-ace0-fbb531ad65f3
|