Dear Bernhard, thanks for the efforts.
Best regards, Tjeerd Pinkert On 1/27/20 15:26, Bernhard Übelacker wrote: > Control: reassign -1 zynaddsubfx-dssi 3.0.3-1 > Control: affects -1 + audacity > > > Hello Tjeerd, > >> Thanks for coming back, I'm not in a hurry... >> >> The problem is that I can not trigger specific bugs (they seem to happen >> somewhat random). So a made some more valgrinds: valgrind.dat > > Because the error des not manifest each run in the same location > this might be thread related e.g. two threads are working on the > same memory. > > >> I had zynaddsubfx installed to try it out and see if I like it, but did >> not uninstall after the tries (sufficient diskspace luxury). I could >> uninstall zynadd, zynsaddsubfx and zynaddsubfx-dssi, the data package is >> depended on by the lmms-common package. > > I tried to install some more lv2 plugins and bridges and after enabling > all in audacity I got the shared library loaded libzynaddsubfx_dssi.so > by just starting audacity - but still cannot reproduce a crash. > (Details attached.) > > >> And audacity runs without crashing (thanks for the hint) but still gives >> a lot of debug output: audacity_debug.dat > > Most of the remaining output seems gui related - seems to be harmless. > > >> So at least the the source of all evil is now clear... and the bug >> "resolved". Do you have contact with the zynaddsubfx devs and file a bug >> there? Devs amongst each other might talk clearer? Otherwise I'm happy >> to do it. > > It might not be that clear - depending on e.g. libzynaddsubfx_dssi.so > is intended to work multi threaded ... > > When I did run audacity in my test environment I get some > "Possible data race" with valgrind's helgrind tool. > > At least it might be related to some plugins, either direct or > via some bridges, therefore I hope it is ok to reassign. > > Bringing the issue upstream might also not be a bad idea. > > Kind regards, > Bernhard
signature.asc
Description: OpenPGP digital signature
