Skip to contentSkip to conversation

: RMS hardware input and software plaback levels are incorrectly reported under 64 bit linux

ALSA - driver 0001225 [email protected] 1 participant 1 message Jul 3, 2005

lore
Share & tools
lore
#0 lore raw

A NOTE has been added to this issue.
======================================================================
<https://bugtrack.alsa-project.org/alsa-bug/view.php?id=1225>
======================================================================

Reported By:                ckg
Assigned To:                charbonnel
======================================================================
Project:                    ALSA - driver
Issue ID:                   1225
Category:                   PCI - RME HDSP
Reproducibility:            always
Severity:                   major
Priority:                   normal
Status:                     assigned
Distribution:               
Kernel Version:             2.6.11
======================================================================
Date Submitted:             07-01-2005 20:11 CEST
Last Modified:              07-03-2005 05:22 CEST

======================================================================
Summary: RMS hardware input and software plaback levels are
incorrectly reported under 64 bit linux
Description:
I'm using the PCI HDSP with a Multiface interface, Everything appears to be correct with the audio, but the hdspmixer reports bogus valus for hardware input and software playback RMS levels. The peak leavels are correct. The problem happens when the driver is compiled into the kernel and when it is a module.

I think the problem is caused because the rms regesters are positioned at a 4 byte but not an 8 byte boundary on a 32 bit system, but on a 64 bit system, they are movd back 4 bytes to an 8 byte alignment
======================================================================

----------------------------------------------------------------------
charbonnel - 07-02-05 08:42
----------------------------------------------------------------------
Hi,

It sounds similar to a bug in the byte swapping code that was fixed some time ago.
(see ALSA BUG https://bugtrack.alsa-project.org/alsa-bug/view.php?id=0000801)
Do you use the in-kernel alsa driver with 2.6.11 ? If so they may be a little old.
What's the version reported by cat /proc/asound/version ?
I suspect your patch seems to fix the problem but introduces a one channel offset in the metering values.

Thomas

----------------------------------------------------------------------
ckg - 07-03-05 05:22
----------------------------------------------------------------------
Thomas,

This sounds like the same problem. cat /proc/asound/version reports that I'm running version 1.0.8. I am using the in-kernal driver.
I took a look at your patch for the problem. I haven't tried it. There seems to be something interesting going on here and I am not able to completely grasp what is happening. Here's the situation:
My patch does not appear to introduce a one channel offset. I have tested extensively and (unless I am seriously missing something) correct levels are reported on the correct inputs.
While debuggin my guess was that running 64-bits caused the port space to be positioned at an 8 byte boundary. I determined that I needed to jump backwards four bytes and everything worked great. but... while debugging I noticed that when I turned the gain up very high on my pre-amp and fed low-level random noise into channel two the hdspmixer would show the rms levels changing on channel one. In fact, as I would turn up the gain on channel two, the rms level would flucuate up and down repeatedly on channel one. My guess was that the least significant bytes of the rms value on channel one was being read into the most significant bytes of the rms for channel one. The trouble is that in order to fix this issue I would need to offset the port addresses by -12, not -4.
If we assume that your fix is correct then it seems quite logical that my fix would introduce a one channel offset. The vise-versa is also true. I need to sleep on this and perhaps something bright will occur to me. In the mean time, both our fixes seem to work, and I wonder if maybe neither of them is entirely correct... Hmmm. Show me what I'm missing.

~ckg

Issue History
Date Modified Username Field Change
======================================================================

07-01-05 20:11 ckg            New Issue                                    
07-01-05 20:11 ckg            File Added: hdsp.patch                       
07-01-05 20:11 ckg            Kernel Version            => 2.6.11          
07-02-05 08:42 charbonnel     Note Added: 0005352                          
07-03-05 05:22 ckg            Note Added: 0005369                          

======================================================================

-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click