Comment 0 for bug 366884

Binary package hint: linux-image-2.6.28-11-generic

PROBLEM:
- Jackd worked for me in Hardy Desktop x86_64 until recent updates (incredibly large xruns - x86-64 jack compile bug?)
- jackd works fine on hardy (2.6.27-14-generic) on eeepc (i386)
- clean Jaunty 9.04 Desktop (x86_64) install on same hardware as the Hardy mentioned:
   - regular and frequent (4-15 per second, depending on jack settings) xruns (~ 0.02 msecs each)
- playing sound of course gives tiny bits (msecs) of sound from the original stream

EXPECTED BEHAVIOR:
- jackd should at least not have xruns when not playing anything

HOW TO REPRODUCE:
- disable/kill pulseaudio
- jackd -d alsa .... (almost any combination of options give the same effect, w/wo realtime, etc)
- this is with _no_ playback - just jackd running

OTHER TESTS:
- same thing on both 2.6.28-11-generic and 2.6.28-3-rt (x86_64) - jackd version: 0.116.1-3ubuntu3
- same thing without X running
- did not check Ubuntu Studio x86_64 yet, nor did I check i386 ...

INFO:
- I used "ubuntu-proposed" repositories for both Hardy and Jaunty
- limits.conf setup correctly for rtprio, memory locking, nice (@audio)
- pulse is installed, but I renamed pulseaudio binary to prevent it from loading

HARDWARE:
- lspci: onboard nVidia MCP51 AC97 Audio Controller (snd_intel8x0)
- /proc/interrupts shows sata_nv and Intel ICH on same irq
- AMD Sempron 2500+, 2GB RAM, 2x SATA HD, onboard GeForce6100

Questions:
- anything else I can attach/test/tryout that would be helpful?
- any specific kernel I should check? (from PPA or build specific version?)

Sorry if this issue is well know or a duplicate of another bug - I couldn't find anything similar or relevant.
The last thing that is left is the latest vanilla kernel and jack sources and try to debug my myself - any suggestions would be helpful, thanks.