ALSA CVS and JACK?

ALSA CVS and JACK?

[Paul Davis]

can jaroslav or takashi fill me in on the current state of ALSA CVS
with respect to JACK? there have been many, many reports of JACK
crashing machines when run with recent ALSA CVS in real-time mode, and
mentions here on alsa-devel of configure-time JACK detection that
causes certain code to be commented out, or something like that.

we are about to do a cross-DAW benchmark of ardour with most
commercial apps, and the people running the test need to install
ALSA. since they will run JACK in RT mode, i need to be sure that this
is not going to crash their system, and would like to be able to
provide assistance to others in this matter.

--p


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: ALSA CVS and JACK?

[Jaroslav Kysela]

On Tue, 25 Feb 2003, Paul Davis wrote:

> can jaroslav or takashi fill me in on the current state of ALSA CVS
> with respect to JACK? there have been many, many reports of JACK
> crashing machines when run with recent ALSA CVS in real-time mode, and
> mentions here on alsa-devel of configure-time JACK detection that
> causes certain code to be commented out, or something like that.

Can we see reports? I don't see any reason for crashes.

						Jaroslav

-----
Jaroslav Kysela <perex@suse.cz>
Linux Kernel Sound Maintainer
ALSA Project, SuSE Labs



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: ALSA CVS and JACK?

[Paul Davis]

>Can we see reports? I don't see any reason for crashes.

nobody that this has happened to so far has had an oops trace. they
are always running graphical apps and do not have a serial console. i
am too nervous (:) to try it myself (hence i still have ALSA CVS from
way back). apparently it may it also be connected to a particular
kernel version.

i would say that at least 6-8 people on jackit-dev have reported this.

--p


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: ALSA CVS and JACK?

[Takashi Iwai]

At Tue, 25 Feb 2003 14:02:05 -0500,
Paul Davis wrote:
> 
> >Can we see reports? I don't see any reason for crashes.
> 
> nobody that this has happened to so far has had an oops trace. they
> are always running graphical apps and do not have a serial console. i
> am too nervous (:) to try it myself (hence i still have ALSA CVS from
> way back). apparently it may it also be connected to a particular
> kernel version.
 
isn't it the 2.4.20+ext3 problem?

> i would say that at least 6-8 people on jackit-dev have reported this.

hmm, i couldn't find the thread...  oops+ALSA hits only the ext3
problem.

could you give the pointer?


Takashi


-------------------------------------------------------
This SF.net email is sponsored by: Scholarships for Techies!
Can't afford IT training? All 2003 ictp students receive scholarships.
Get hands-on training in Microsoft, Cisco, Sun, Linux/UNIX, and more.
www.ictp.com/training/sourceforge.asp

Re: ALSA CVS and JACK?

[jfm3]

On Wed, 2003-02-26 at 04:13, Takashi Iwai wrote:
> At Tue, 25 Feb 2003 14:02:05 -0500,
> Paul Davis wrote:
> > 
> > >Can we see reports? I don't see any reason for crashes.
> > 
> > nobody that this has happened to so far has had an oops trace. they
> > are always running graphical apps and do not have a serial console. i
> > am too nervous (:) to try it myself (hence i still have ALSA CVS from
> > way back). apparently it may it also be connected to a particular
> > kernel version.
>  
> isn't it the 2.4.20+ext3 problem?
> 

OK, I switched to ext2 on all file systems. Now I can run latencytest to
completion, although oddly it complains about how it can't kill top. The
upper bound for audio latency on this crusoe-based c1mw is presently
6.8ms, both according to latencytest and the pd latency measuring patch.
The cpu latency is reported at around 1.5ms.

Jack and freqtweak still make the system hang, although I have to try
several times to start freqtweak at all without getting the error
message that freqtweak was "shut down" by the jack server.

-- 
(jfm3  2838 BCBA 93BA 3058 ED95  A42C 37DB 66D1 B43C 9FD0)



-------------------------------------------------------
This SF.net email is sponsored by: Scholarships for Techies!
Can't afford IT training? All 2003 ictp students receive scholarships.
Get hands-on training in Microsoft, Cisco, Sun, Linux/UNIX, and more.
www.ictp.com/training/sourceforge.asp

Re: ALSA CVS and JACK?

[Maarten de Boer]

> can jaroslav or takashi fill me in on the current state of ALSA CVS
> with respect to JACK? there have been many, many reports of JACK
> crashing machines when run with recent ALSA CVS in real-time mode, and
> mentions here on alsa-devel of configure-time JACK detection that
> causes certain code to be commented out, or something like that.

I don't find those mentions, and I don't find the code either. (Though
my CVS is not 100% uptodate). I believe that the only configure-time
JACK detection is related to the pcm_jack plugin. If jack is not present
it should not be compiled, because it will cause link errors. 

I also have experienced crashed (well, freezes) with jack in RT mode,
but I don't think it has to do anything with alsa.

Maarten


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: ALSA CVS and JACK?

[Paul Davis]

>> can jaroslav or takashi fill me in on the current state of ALSA CVS
>> with respect to JACK? there have been many, many reports of JACK
>> crashing machines when run with recent ALSA CVS in real-time mode, and
>> mentions here on alsa-devel of configure-time JACK detection that
>> causes certain code to be commented out, or something like that.
>
>I don't find those mentions, and I don't find the code either. 

you're right. i misread the message i was thinking of. but i am sure
jaroslav said something about not understanding why a certain piece of
code was a problem with/without JACK.

>I also have experienced crashed (well, freezes) with jack in RT mode,
>but I don't think it has to do anything with alsa.

all the evidence we have on jackit-dev points to this. people can
switch between ALSA versions, and it crashes with a newer copy and not
with an earlier one. there are some suggestions that its a kernel+ALSA
problem. the fact that it only happens in RT mode suggests some
interaction of SCHED_FIFO scheduling and kernel locking or something
like that.

--p


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: ALSA CVS and JACK?

[jfm3]

On Tue, 2003-02-25 at 14:00, Paul Davis wrote:
> >> can jaroslav or takashi fill me in on the current state of ALSA CVS
> >> with respect to JACK? there have been many, many reports of JACK
> >> crashing machines when run with recent ALSA CVS in real-time mode, and
> >> mentions here on alsa-devel of configure-time JACK detection that
> >> causes certain code to be commented out, or something like that.
> >
> >I don't find those mentions, and I don't find the code either. 

I have this problem on my crusoe laptop. I am using kernel
2.4.20+pre+ll, alsa with "cvsdate 20030218.141220", and jack 0.51.1.
kernel and alsa are built by me, jack is from Planet CCRMA. I have no
serial port on this laptop, just a USB/serial converter, but I will be
happy to run whatever test anyone might describe to help find the
problem.

-- 
(jfm3  2838 BCBA 93BA 3058 ED95  A42C 37DB 66D1 B43C 9FD0)



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf