Red Hat Bugzilla – Bug 326871
zynaddsubfx improper blocking call in Jack process function
Last modified: 2007-11-30 17:12:18 EST
Description of problem:
ZynAddSubFX was not very well designed to work with the Jack audio server, the
jack process call can block on a mutex (it should not) and then the jack client
delays processing the whole jack graph. This patch by Lars Luthman avoids the
problem and zynaddsubfx becomes a lot more reliable, specially with the current
version of svn jack which is much more sensitive to hiccups in the clients.
Version-Release number of selected component (if applicable):
Current version is 2.2.1-14.fcx
Steps to Reproduce:
1. For example, change the current patch, you should observe a glitch in the
Jack output. If the glitch is long enough ZynAddSubFX gets kicked out of the
Jack processing graph.
Xruns in the Jack audio server
No xruns in the Jack audio server
Patch originally by Lars Luthman, tweaked to patch on zyn with no preexisting
Jack midi patches.
Attached files: zynaddsubfx-unzombify.patch
Created attachment 223291 [details]
Created attachment 223301 [details]
spec file changes
Thanks. I agree that ZynAddSubFX has problems in this area. I looked into it
about a year ago with an eye to have it use lockless buffers. I'll push out a
new zyn for devel, F7 and FC-6.
(In reply to comment #3)
> Thanks. I agree that ZynAddSubFX has problems in this area. I looked into it
> about a year ago with an eye to have it use lockless buffers. I'll push out a
> new zyn for devel, F7 and FC-6.
Thanks! Obviously it'd be good to have a proper fix but this patch improves the
situation a lot. I'm testing a new svn jack and zyn without this is not really
I've kicked off builds for FE-6, F-7 and devel. The next updates should include
this patch. I'm closing this bug. Thanks nando!