[Jack-Devel] JACK1 0.125.0rc1 now available

classic Classic list List threaded Threaded
14 messages Options
Reply | Threaded
Open this post in threaded view
|

[Jack-Devel] JACK1 0.125.0rc1 now available

Paul Davis
There's a new release candidate of JACK1 now available. Because of the magnitude of a couple of the changes, some wider testing would be appreciated before 0.125.0 can be released.


(tagged in git as 0.125.0.rc1)

Changes since 0.124.1, in rough order of significance:

   * use topological sort when ordering clients. Previous sorting implementation would
     generate incorrect ordering of clients in some (unusual, but not uncommon) conditions.
   * in the alsa_midi slave driver, fix hotpug device enumeration. JACK1 will now respond
     to MIDI devices being connected and disconnected while running.
   * increase maximum size of a single JACK MIDI event to 64 bytes
   * in the alsa_midi slave driver, fix thread start/stop handling when freewheeling
   * add support for jack_port_rename()
   * drop support for CPU cycle counting clock, use kernel clocksource instead
   * fix a double-fork that left zombie processes around
   * improve the validity and usability of the return value of jack_frame_time()
   * fix failing metadata look up by clearing UUID parsing buffer before use
   * support unescaped double quotes in $HOME/.jackdrc
   * fix memory leaks of metata key/value pairs
   * fix crash caused by incorrect jack_error() format string
   * remove option help from jackd and point user at documentation
   * fix problems with garbage keys in metadata
   * fix out-of-tree builds
   * update configure.ac to work with current-ish autotools with less errors
   * use gcc atomics and CLOCK_REALTIME for generic CPU builds
   * a handful of other minor bug fixes

Contributors: Hans-Peter Portner, Fons Adriaensen, Erik de Castro Lopo, Filipe Coehlo, David Robillard, Adrian Knoth,  Dominic Sacré, Peter Nelson, Rui Nuno Capela, Robin Gareus, Peter Nelson, Paul Davis



_______________________________________________
Jack-Devel mailing list
[hidden email]
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org
Reply | Threaded
Open this post in threaded view
|

Re: JACK1 0.125.0rc1 now available

Tim-2
On February 23, 2016 11:42:25 AM Paul Davis wrote:

> There's a new release candidate of JACK1 now available. Because of the
> magnitude of a couple of the changes, some wider testing would be
> appreciated before 0.125.0 can be released.
>
>
> http://jackaudio.org/downloads/jack-audio-connection-kit-0.125.0rc1.tar.gz
>
> (tagged in git as 0.125.0.rc1)
>
> Changes since 0.124.1, in rough order of significance:
>
>    * use topological sort when ordering clients. Previous sorting
> implementation would
>      generate incorrect ordering of clients in some (unusual, but not
> uncommon) conditions.
>    * in the alsa_midi slave driver, fix hotpug device enumeration. JACK1
> will now respond
>      to MIDI devices being connected and disconnected while running.
>    * increase maximum size of a single JACK MIDI event to 64 bytes
>    * in the alsa_midi slave driver, fix thread start/stop handling when
> freewheeling
>    * add support for jack_port_rename()
>    * drop support for CPU cycle counting clock, use kernel clocksource
> instead
>    * fix a double-fork that left zombie processes around
>    * improve the validity and usability of the return value of
> jack_frame_time()
>    * fix failing metadata look up by clearing UUID parsing buffer before use
> * support unescaped double quotes in $HOME/.jackdrc
>    * fix memory leaks of metata key/value pairs
>    * fix crash caused by incorrect jack_error() format string
>    * remove option help from jackd and point user at documentation
>    * fix problems with garbage keys in metadata
>    * fix out-of-tree builds
>    * update configure.ac to work with current-ish autotools with less errors
> * use gcc atomics and CLOCK_REALTIME for generic CPU builds
>    * a handful of other minor bug fixes
>
> Contributors: Hans-Peter Portner, Fons Adriaensen, Erik de Castro Lopo,
> Filipe Coehlo, David Robillard, Adrian Knoth,  Dominic Sacré, Peter Nelson,
> Rui Nuno Capela, Robin Gareus, Peter Nelson, Paul Davis
Haven't tried building yet, but I noticed a fix for forcing configuration
 was added. Thanks.

But I am wondering if the second half of my submitted patch is required.
It fixes missing code in jack_port_by_name and jack_port_by_name_int.

Attached file. Have a look.

Thanks.
Tim.

_______________________________________________
Jack-Devel mailing list
[hidden email]
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

jack1_port_by_name_fix_ver_2.diff (4K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: JACK1 0.125.0rc1 now available

Erik de Castro Lopo-10
In reply to this post by Paul Davis
Paul Davis wrote:

> There's a new release candidate of JACK1 now available. Because of the
> magnitude of a couple of the changes, some wider testing would be
> appreciated before 0.125.0 can be released.

There are still a couple of open PRs here:

    https://github.com/jackaudio/tools/pulls
    https://github.com/jackaudio/example-clients/pulls

Erik
--
----------------------------------------------------------------------
Erik de Castro Lopo
http://www.mega-nerd.com/
_______________________________________________
Jack-Devel mailing list
[hidden email]
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org
Reply | Threaded
Open this post in threaded view
|

Re: JACK1 0.125.0rc1 now available

Paul Davis
There's also a significant regression since 0.124.X. I am having problems with clients being zombified AND with jackd crashing. Work on this issue is slow and will delay release.

But thanks for the heads up/reminder.

On Sat, Feb 27, 2016 at 9:32 PM, Erik de Castro Lopo <[hidden email]> wrote:
Paul Davis wrote:

> There's a new release candidate of JACK1 now available. Because of the
> magnitude of a couple of the changes, some wider testing would be
> appreciated before 0.125.0 can be released.

There are still a couple of open PRs here:

    https://github.com/jackaudio/tools/pulls
    https://github.com/jackaudio/example-clients/pulls

Erik
--
----------------------------------------------------------------------
Erik de Castro Lopo
http://www.mega-nerd.com/
_______________________________________________
Jack-Devel mailing list
[hidden email]
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org


_______________________________________________
Jack-Devel mailing list
[hidden email]
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org
Reply | Threaded
Open this post in threaded view
|

Re: JACK1 0.125.0rc1 now available

Erik de Castro Lopo-10
Paul Davis wrote:

> There's also a significant regression since 0.124.X. I am having problems
> with clients being zombified AND with jackd crashing. Work on this issue is
> slow and will delay release.

Is this issue documented anyway? Even just steps required to reproduce it.
Anything anyone else can do to help?

Cheers,
Erik
--
----------------------------------------------------------------------
Erik de Castro Lopo
http://www.mega-nerd.com/
_______________________________________________
Jack-Devel mailing list
[hidden email]
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org
Reply | Threaded
Open this post in threaded view
|

Re: JACK1 0.125.0rc1 now available

Paul Davis


On Sat, Feb 27, 2016 at 9:45 PM, Erik de Castro Lopo <[hidden email]> wrote:
Paul Davis wrote:

> There's also a significant regression since 0.124.X. I am having problems
> with clients being zombified AND with jackd crashing. Work on this issue is
> slow and will delay release.

Is this issue documented anyway? Even just steps required to reproduce it.
Anything anyone else can do to help?

I don't have a recipe. Just my normal daytime activity (developing ardour, playing music, listening to streaming music, watching videos) now kills JACK several times a day, whereas it used to be rock solid.

I have not tried git bisect yet, but I will when I get a moment.
 

_______________________________________________
Jack-Devel mailing list
[hidden email]
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org
Reply | Threaded
Open this post in threaded view
|

Re: JACK1 0.125.0rc1 now available

Rui Nuno Capela
On 02/28/2016 03:04 AM, Paul Davis wrote:

>
>
>> There's also a significant regression since 0.124.X. I am having problems
>> with clients being zombified AND with jackd crashing. Work on this issue is
>> slow and will delay release.
>
> Is this issue documented anyway? Even just steps required to reproduce it.
> Anything anyone else can do to help?
>
>
> I don't have a recipe. Just my normal daytime activity (developing
> ardour, playing music, listening to streaming music, watching videos)
> now kills JACK several times a day, whereas it used to be rock solid.
>
> I have not tried git bisect yet, but I will when I get a moment.
>

fwiw. have built, packaged and run jack-0.125.0rc1:
   http://www.rncbc.org/archive/#jack
also on my OBS home project:
   https://build.opensuse.org/package/show/home:rncbc/jack

example startup log head:
~~~
/usr/bin/jackd -P86 -t3000 -m -dalsa -dhw:DMX6Fire -r48000 -p128 -n2 -Xseq
jackd 0.125.0rc1
Copyright 2001-2009 Paul Davis, Stephane Letz, Jack O'Quinn, Torben Hohn
and others.
jackd comes with ABSOLUTELY NO WARRANTY
This is free software, and you are welcome to redistribute it
under certain conditions; see the file COPYING for details
JACK compiled with System V SHM support.
loading driver ..
apparent rate = 48000
creating alsa driver ...
hw:DMX6Fire|hw:DMX6Fire|128|2|48000|0|0|nomon|swmeter|-|32bit
configuring for 48000Hz, period = 128 frames (2.7 ms), buffer = 2 periods
ALSA: final selected sample format for capture: 32bit integer little-endian
ALSA: use 2 periods for capture
ALSA: final selected sample format for playback: 32bit integer little-endian
ALSA: use 2 periods for playback
creating alsa_midi driver ...


using qjackctl for that matter where pretty_name metadata  aliases are
being used all the time btw. also doing all the usual things:
development, testing, debugging, playing, streaming, browsing, etc. and
with pulseaudio-sink and -source modules all along... all that on one
desktop and two laptops, all rigged with latest 4.4 kernels (preempt
w/threadirqs and preempt_rt_full; rtirq sets the primary snd card to
rtprio=90 thus the jackd -P86 ... ;))

all to tell that 0.125.0rc1 is doing fine over here without any major
hiccups, though the usual xruns specially due on video and possibly
pulseaudio related but sure that was also happening before on jackd-0.124.x

so it seems it's all as solid as ever before.

however i have a complaint :) re. the git repo in particular: since
0.125.0rc1 announcement the submodules aren't pulling right... evidence
follows:

~~~
~/src/jack1/jack1-git> git pull --all
Fetching origin
Already up-to-date.

~/src/jack1/jack1-git> git submodule update
fatal: reference is not a tree: e74323c535876abca9a2293bd2ca426a3b91ff8b
fatal: reference is not a tree: a2536c8b348bbfb21ec53db901e7992d7ffef474
fatal: reference is not a tree: 8e140b72de0231d129c6006db969f1dba4f1486b
Unable to checkout 'e74323c535876abca9a2293bd2ca426a3b91ff8b' in
submodule path 'example-clients'
Unable to checkout 'a2536c8b348bbfb21ec53db901e7992d7ffef474' in
submodule path 'jack'
Unable to checkout '8e140b72de0231d129c6006db969f1dba4f1486b' in
submodule path 'tools'

~/src/jack1/jack1-git> git remote -v
origin  git://github.com/jackaudio/jack1.git (fetch)
origin  git://github.com/jackaudio/jack1.git (push)

~/src/jack1/jack1-git> git branch -a
* master
remotes/origin/HEAD -> origin/master
remotes/origin/control_device_name_fixes
remotes/origin/master
remotes/origin/uuid


byee
--
rncbc aka. Rui Nuno Capela
_______________________________________________
Jack-Devel mailing list
[hidden email]
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org
Reply | Threaded
Open this post in threaded view
|

Re: JACK1 0.125.0rc1 now available

Rui Nuno Capela
On 02/28/2016 10:49 AM, Rui Nuno Capela wrote:
>
> ... 0.125.0rc1 is doing fine over here without any major hiccups,
> though the usual xruns specially due on video and possibly
> pulseaudio related but sure that was also happening before on
> jackd-0.124.x
>
> so it seems it's all as solid as ever before.
>

not so fast, sorry. i missed to report one particular issue though:

as a matter of fact, the alsa_midi driver eg. jackd -Xalsa_midi (or
-Xseq), seems to severely interfere with pulseaudio and/or vice-versa.
  symptoms are that of all jack audio pcm sound getting horribly garbled
if not silence most of the time and pulseaudio stops to function if at
all. nb. though good old `a2jmidid -e &` works fine instead.

however, i am not sure this is incidental to 0.125.0rc1, as i believe
the issue has been seen before and probably is related to some
pulseaudio recent updates. btw. distro is opensuse tumbleweed which gets
pervasive system updates, on a weekly basis

again, to be clear, nothing about zombie clients or crashes experienced
with 0.125.0rc1.

seeya
--
rncbc aka. Rui Nuno Capela
_______________________________________________
Jack-Devel mailing list
[hidden email]
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org
Reply | Threaded
Open this post in threaded view
|

Re: JACK1 0.125.0rc1 now available

Hanspeter Portner
In reply to this post by Paul Davis
On 28.02.2016 04:04, Paul Davis wrote:

>
>
> On Sat, Feb 27, 2016 at 9:45 PM, Erik de Castro Lopo <[hidden email] <mailto:[hidden email]>> wrote:
>
>     Paul Davis wrote:
>
>     > There's also a significant regression since 0.124.X. I am having problems
>     > with clients being zombified AND with jackd crashing. Work on this issue is
>     > slow and will delay release.
>
>     Is this issue documented anyway? Even just steps required to reproduce it.
>     Anything anyone else can do to help?
>
>
> I don't have a recipe. Just my normal daytime activity (developing ardour, playing music, listening to streaming music,
> watching videos) now kills JACK several times a day, whereas it used to be rock solid.
>
> I have not tried git bisect yet, but I will when I get a moment.

I have seen issues, too. I get heavily crackled sound on a first clients audio output upon loading of a second client
into the graph.
The crackling sometimes disappears when loading a third client.
The server may also crash sometimes while crackling along.

I have bisected it down to: 423931219dd3e3b669fde97786cadae92c066dc1
before that I do not seem to experience these crackles.

.jackdrc
--------
/usr/bin/jackd -M2048 -Xalsa_midi -P70 -t2000 -dalsa -dhw:Juli -r48000 -p128 -n3
_______________________________________________
Jack-Devel mailing list
[hidden email]
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org
Reply | Threaded
Open this post in threaded view
|

Re: JACK1 0.125.0rc1 now available

Paul Davis


On Wed, Mar 2, 2016 at 4:27 PM, Hanspeter Portner <[hidden email]> wrote:
On 28.02.2016 04:04, Paul Davis wrote:
>
>
> On Sat, Feb 27, 2016 at 9:45 PM, Erik de Castro Lopo <[hidden email] <mailto:[hidden email]>> wrote:
>
>     Paul Davis wrote:
>
>     > There's also a significant regression since 0.124.X. I am having problems
>     > with clients being zombified AND with jackd crashing. Work on this issue is
>     > slow and will delay release.
>
>     Is this issue documented anyway? Even just steps required to reproduce it.
>     Anything anyone else can do to help?
>
>
> I don't have a recipe. Just my normal daytime activity (developing ardour, playing music, listening to streaming music,
> watching videos) now kills JACK several times a day, whereas it used to be rock solid.
>
> I have not tried git bisect yet, but I will when I get a moment.

I have seen issues, too. I get heavily crackled sound on a first clients audio output upon loading of a second client
into the graph.
The crackling sometimes disappears when loading a third client.
The server may also crash sometimes while crackling along.

I have bisected it down to: 423931219dd3e3b669fde97786cadae92c066dc1
before that I do not seem to experience these crackles.


I get the crackles *sometimes*. My general setup is Clementine playing audio, then ardour starts up. 1 time in 10, Clementine will crackle like crazy and then it will die and/or the server will die. Not using Clementine, 1 in 15 times that I run Ardour, I can kill the server.

I also bisected it down to fons' topological sort patch. Which is a problem because in some ways that's the biggest reason for this new release :(
 

_______________________________________________
Jack-Devel mailing list
[hidden email]
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org
Reply | Threaded
Open this post in threaded view
|

Re: JACK1 0.125.0rc1 now available

Rui Nuno Capela
On 03/02/2016 09:46 PM, Paul Davis wrote:

> On Wed, Mar 2, 2016 at 4:27 PM, Hanspeter Portner
> <[hidden email]
>
>> I have seen issues, too. I get heavily crackled sound on a first
>> clients audio output upon loading of a second client into the
>> graph. The crackling sometimes disappears when loading a third
>> client. The server may also crash sometimes while crackling along.
>>
>> I have bisected it down to:
>> 423931219dd3e3b669fde97786cadae92c066dc1 before that I do not seem
>> to experience these crackles.
>
> I get the crackles *sometimes*. My general setup is Clementine
> playing audio, then ardour starts up. 1 time in 10, Clementine will
> crackle like crazy and then it will die and/or the server will die.
> Not using Clementine, 1 in 15 times that I run Ardour, I can kill the
> server.
>
> I also bisected it down to fons' topological sort patch. Which is a
> problem because in some ways that's the biggest reason for this new
> release :(
>

i must say that i can also now definitely confirm that the issue below,
that i've reported earlier as garbled sound on pulseaudio+alsa_midi
driver, is introduced by this very same "Fons' topological sort" commit.

only that the "crackling" does occur *always* on me (if not dead silence).

funny enough, in my case at least, on all tests i've tried, and they
were a lot during the past week, it only occurred deterministically
speaking, when pulseaudio-jack-sink and -source modules were up and
jackd had the alsa_midi slave driver loaded (via -Xalsa_midi or -Xseq).

if the alsa_midi slave driver is not loaded, everything runs and sounds
fine (apparently).

also, using a2jmidid as a backup alsa_midi bridge, the problem didn't
manifest.

oh, and the new release has way more important fixes to me at least,
like the fixes on the alsa_midi slave driver hotplug and freewheel
handling and the metadata memory leaks and garbage/crash fixes. etc.


On 02/28/2016 12:24 PM, Rui Nuno Capela wrote:
 >
 > not so fast, sorry. i missed to report one particular issue though:
 >
 > as a matter of fact, the alsa_midi driver eg. jackd -Xalsa_midi (or
 > -Xseq), seems to severely interfere with pulseaudio and/or
 > vice-versa. symptoms are that of all jack audio pcm sound getting
 > horribly garbled if not silence most of the time and pulseaudio stops
 > to function if at all. nb. though good old `a2jmidid -e &` works fine
 > instead.
 >
 > however, i am not sure this is incidental to 0.125.0rc1, as i
 > believe the issue has been seen before and probably is related to
 > some pulseaudio recent updates. btw. distro is opensuse tumbleweed
 > which gets pervasive system updates, on a weekly basis
 >
 > again, to be clear, nothing about zombie clients or crashes
 > experienced with 0.125.0rc1.
 >

hth.
byee
--
rncbc aka. Rui Nuno Capela
_______________________________________________
Jack-Devel mailing list
[hidden email]
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org
Reply | Threaded
Open this post in threaded view
|

Re: JACK1 0.125.0rc1 now available

Erik de Castro Lopo-10
In reply to this post by Paul Davis
Paul Davis wrote:

> I also bisected it down to fons' topological sort patch. Which is a problem
> because in some ways that's the biggest reason for this new release :(

Any clues as to why this topological sort code behaves different from the
code it replaces? Does have/not have locking? Does it have a worse runtime
complexity? Does it have a bug?

Just looking for clues.

Erik
--
----------------------------------------------------------------------
Erik de Castro Lopo
http://www.mega-nerd.com/
_______________________________________________
Jack-Devel mailing list
[hidden email]
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org
Reply | Threaded
Open this post in threaded view
|

Re: JACK1 0.125.0rc1 now available

Paul Davis
I didn't bother to read Fons' code in much detail ... I was more concerned with how to reformat it so as to get a "clean" commit. Perhaps somebody else will read that commit and be able to see what the issue(s) are.

On Mon, Mar 7, 2016 at 1:55 AM, Erik de Castro Lopo <[hidden email]> wrote:
Paul Davis wrote:

> I also bisected it down to fons' topological sort patch. Which is a problem
> because in some ways that's the biggest reason for this new release :(

Any clues as to why this topological sort code behaves different from the
code it replaces? Does have/not have locking? Does it have a worse runtime
complexity? Does it have a bug?

Just looking for clues.

Erik
--
----------------------------------------------------------------------
Erik de Castro Lopo
http://www.mega-nerd.com/
_______________________________________________
Jack-Devel mailing list
[hidden email]
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org


_______________________________________________
Jack-Devel mailing list
[hidden email]
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org
Reply | Threaded
Open this post in threaded view
|

Re: JACK1 0.125.0rc1 now available

Thomas Brand
In reply to this post by Erik de Castro Lopo-10
On Mon, March 7, 2016 07:55, Erik de Castro Lopo wrote:

> Paul Davis wrote:
>
>
>> I also bisected it down to fons' topological sort patch. Which is a
>> problem because in some ways that's the biggest reason for this new
>> release :(
>
> Any clues as to why this topological sort code behaves different from the
>  code it replaces? Does have/not have locking? Does it have a worse
> runtime complexity? Does it have a bug?
>
> Just looking for clues.
>
>
> Erik
> --

for all practical purposes it would be strange if the new sort patch would
behave the same as without the patch. the question is how it behaves
differently and the possible consequences it might impose on clients. i'm
sure fons did some tests.
apropos tests, a set of jack1/jack2 generic clients, setup in a series or
in parallel and configurable client "behaviour" in terms of cycle time
used, cpu .. could help to spot the differencies with numbers.
greetings
Thomas

_______________________________________________
Jack-Devel mailing list
[hidden email]
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org