18 Jun
2006
18 Jun
'06
2:33 p.m.
On 6/17/06, Lee Revell <rlrevell(a)joe-job.com> wrote:
It's probably a sound driver bug. This HDA intel crap is really getting
to be a nightmare. It seems like sound is broken on every other new
laptop.
Strange, Intel is no nice with OSS video drivers..
Can you try the -rt kernel and enable latency tracing?
You mean compile a custom kernel? Sure, I'll get back once I've tried it.
Does it make any difference if you boot with ACPI
disabled?
No apparent change.
Check /proc/interrupts for your soundcard - when you launch JACK in
realtime mode does the interrupt count increase?
I *think* so, I'm not sure what the format of /proc/interrupts is.
Here's what I saw:
*jack not started*
11: 852 XT-PIC uhci_hcd:usb2, HDA Intel
*started jack*
11: 1010 XT-PIC uhci_hcd:usb2, HDA Intel
11: 1106 XT-PIC uhci_hcd:usb2, HDA Intel
11: 1325 XT-PIC uhci_hcd:usb2, HDA Intel
11: 1537 XT-PIC uhci_hcd:usb2, HDA Intel
11: 1591 XT-PIC uhci_hcd:usb2, HDA Intel
*jack crashed*
11: 1591 XT-PIC uhci_hcd:usb2, HDA Intel
Alright, I'll go try a custom kernel. Ugh, I was really hoping I could
go with packaged kernels.
Lee
Thanks!
On 6/17/06, Jack O'Quin <jack.oquin(a)gmail.com> wrote:
On 6/14/06, I. E. Smith-Heisters
<public(a)0x09.com> wrote:
Okay, looked at it some more. When RT is enabled,
jack just locks up
and the watchdog terminates the process, regardless of the buffer
size. When RT is disabled the xruns are allowed to continue, and the
number of xruns decreases with a higher buffer size (but never go
below about 10/second). There's no evidence that RT mode has failed to
be set. This is all as root.
I am using the proprietary NVIDIA drivers, as gotten from the Ubuntu
repositories. I would be surprised if this had anything to do with it
though, since direct alsa works fine with the same xOrg drivers.
Unless, of course, there's some software conflict between the video
drivers and jack itself (as opposed to there being a hardware-level
conflict).
It would not surprise me for the proprietary drivers to behave in
a non-realtime-safe manner. This would affect JACK much worse
than some heavily-buffered ALSA application.
Can you try it with the open source driver to compare?
--
joq