Red Hat Bugzilla – Bug 81209
(SCHEDULER ?)Sound distortion on VIA 686b
Last modified: 2013-07-02 22:09:53 EDT
Description of problem:
This is one of the bug reports that are unlikely to get fixed,
Steps to Reproduce:
1. install gnome, redhat-artwork, ... (HW described below)
2. ogg123 /usr/share/redhat-credits.ogg
3. Ctrl-Alt-Shift-triple-click the desktop
Expected results: (8.0)
The sound is same in both cases, including volume and the silence at start.
In step 3), I hear clicking sounds at start (seem to be synchronized
with desktop redrawing). When the real music starts, it is
very distorted and LOUD.
lspci gives VIA VT82C656 rev. 50, subsystem ID: ECS, unknown product 0987
Clicking may be X traffic. Heavy video traffic on some boxes causes that.
Playing with BIOS PCI settings can sometimes help as can setting card specific
options in X.
The volume one isnt the kernel - that sounds like something is touching mixer
settings in the
Sorry, I forgot to say explicitly that in RHL 8.0 there is no such problem, and redhat-credits apparently hasn't changed.
Bug currently seems to be on kernel but assigned to me, I don't know
its history but current situation makes no sense so reassigning.
Still the same in final.
Strange thing is that starting
ogg123 /usr/share/redhat-credigs.ogg & /usr/libexec/redhat-credits
(or starting ogg123 longer before starting redhat-credits)
results in perfect sound.
Looks like an O(1) scheduler interaction example
I think this behavior became far worse with recent (1995+) rawhide kernels,
although I am not entirely sure if this is the same behavior that I am seeing.
FWIW, neither of these helped:
* renicing ogg123 to -20
* renicing ogg123 and esd to -20
* the above + renicing redhat-credits to +19
* the above + renicing X to +19
Swhat is a sound card supposed to do if it doesn't get data?
Is it supposed to
* be silent
* repeat last sample
* output random noise
What I actually get is quite different from (cat /dev/urandom > /dev/dsp),
(most noticeable are regular clicks), but it is certainly not similar to
the silence at the start of the ogg file.
Bug in redhat-credits actually, causing ogg123 to dump its real-time console
output to /dev/dsp.
*** This bug has been marked as a duplicate of 100918 ***
Changed to 'CLOSED' state since 'RESOLVED' has been deprecated.