Bug 24354

Summary:	Sound gets badly interrupted ormutilated (hard to describe) if autospeedstep is or was running
Product:	Gentoo Linux	Reporter:	Mark Redel <info>
Component:	[OLD] Unspecified	Assignee:	Tom Payne (RETIRED) <twp>
Status:	RESOLVED CANTFIX
Severity:	normal	CC:	sound, x86-kernel
Priority:	High
Version:	unspecified
Hardware:	x86
OS:	Linux
Whiteboard:
Package list:		Runtime testing required:	---

Description Mark Redel 2003-07-12 14:21:17 UTC

I installed autospeedstep by using the .ebuid-file becaus it was masked (but I
found no bug report on bugzilla).
It compiled well and is functioning ok but after a while if I use xmms or xine
the sound output gets multilated. Killing autospeedstep has no effect to this
also  I change the speed by hand.
Changing speed only by hand is functioning pretty good.
On xmms i can temporarely bypass the malfunction by pause and replay the title
which is played at that moment - on xine i have to stop the film and play it
again (but this is only a temporary solution - after a while it occurs again)

Reproducible: Always
Steps to Reproduce:
1. /etc/init.d/autospeedstep start
2. using xine or xmms and play sound files 
3. it occurs at undefined time (randomly)

Actual Results:  
see obove (details)

Expected Results:  
the software should only change the speedstep of my mobil cpu by using
/proc/acpi/processor/CPU0/performance

I have a toshiba satellite pro 2100 (P4m 2MHz 1GB RAM) Notebook with gentoo:
[ebuild   R  ] media-sound/alsa-driver-0.9.2
sys-apps/autospeedstep [ Masked ] : Latest version installed: 0.2
[ebuild   R  ] media-sound/xmms-1.2.7-r20
[ebuild   R  ] media-video/xine-ui-0.9.20
[ebuild   R  ] media-libs/xine-lib-1_beta10
Linux version 2.4.22_pre2-gss (root@ws01.localnet) (gcc version 3.2.2) #1 Wed
Jul 9 15:39:36 CEST 2003
Problem is also repuroduceable with Kernel 2.4.21-rc8-gss

Comment 1 J. Ellis (RETIRED) gentoo-dev

2003-07-18 23:44:05 UTC

Issue with autospeedstep ebuild not sound.

Comment 2 Tom Payne (RETIRED) gentoo-dev

2003-08-10 02:23:15 UTC

autospeedstep is a bit of a hack: it just monitors the cpu load and adjusts the CPU freq via ACPI accordingly. I doubt the sound driver is clever enough to detect these changes and consequently timing gets mixed up.

I'd recommend using the CPU freq stuff in more recent kernels with the cpufreqd program. This might solve your problem.

Sorry I can't be of more help. Reopen the bug if you want me to look into it more.

Comment 3 Tom Payne (RETIRED) gentoo-dev

2003-09-25 17:55:12 UTC

Resolution lost by bugzilla

Comment 4 Tom Payne (RETIRED) gentoo-dev

2003-09-25 17:55:34 UTC

oops

Comment 5 Tom Payne (RETIRED) gentoo-dev

2003-09-27 19:38:05 UTC

Resolution should be CANTFIX