See below (from the linux-audio-user list):
Re: [LAU] question about jackdmp's ability to detect its soundcard has disconnected
Could this be the difference between 1.9.8 and git version?
On Oct 29, 2012 12:10 PM, "Robin Gareus" robin@gareus.org wrote:
On 10/29/2012 03:55 PM, Ivica Ico Bukvic wrote:
All,
I noticed since upgrading to Ubuntu 12.04 that the default Jack server is jackdmp (v.1.9.8). Please correct me if I am wrong but my understanding
this
is the smp-enabled version 2 that is being developed in parallel with 1.x series of regular jack.
correct. They are also API and ABI compatible.
Either way, one of my concerns is that while in the old jack when one accidentally pulled the soundcard jack was trying to talk to (e.g. a USB soundcard), the jack would gracefully stop.
In my experience it stopped but never gracefully (ie it crashed). but it's been a while since I used jack1.
The new version (and this could be Ubuntu quirk) instead of stopping is pegged in some kind of a spinlock that often times locks up the machine (this may be in part since I am running a lowlatency kernel with audio group given priority) and at times keeps it operational while hogging the cpu and making the computer barely responsive until jackd is killed. Apps connected to jack that I tested so far are also stuck in a loop waiting for a response from jack (although
this
could be the app's shortcoming)--jack never broadcasts a signal that it
has
lost the soundcard (AFAICT). FWIW I am running jackdmp through qjackctl... Any thoughts on this matter would be most appreciated.
I can't reproduce this. jack2 stops the backend when the device is disconnected. You can switch to a new backend using jack_control (sending some Dbus messages to jackd) without re-starting jackd. No CPU hogging here. - I'm running jack2 - aka 1.9.9.4 from git.
robin