[dmasound] Power Computing DEAD dbdma fix (finally)

[dmasound] Power Computing DEAD dbdma fix (finally)

[Iain Sandoe]

[-- Attachment #1: Type: text/plain, Size: 826 bytes --]

Hi List,

Here is (finally working, we think) a patch for people with Power Computing
machines which exhibit DEAD dbdma status on PCI/dma interaction.

This patch is for test - the driver ID comes up 99.7 - and please ignore the
comment about spin-locks.  I've left those out of the patch because they are
still under test.

Many thanks to Michael Zucca who tested the latest set of changes and Kostas
Gewrgiou who did most of the orignally trials...

Please note that it cannot cure the "glitch" - by the time dbdma is DEAD the
damage is done - what it should do is to keep the sound going with minimum
disruption

The patch should apply to either BK 2.4.x  (after my other patch has been
applied) or to Ben H's tree (which already has those patches).

enjoy - and Power Computing machine owners - tell us how it does.

Iain.

[-- Attachment #2: dmasound-power-comp.diff --]
[-- Type: audio/x-mpeg, Size: 5441 bytes --]

Re: [dmasound] Power Computing DEAD dbdma fix (finally)

[Michael R. Zucca]

Hats off to Iain for fixing this bug!

The fix can't be understated. I noticed that when this patch wasn't fixed,
the DEAD dbdma engine was somehow messing with PCI bus data. I noticed this
under X when I was using sound, there was garbage in some of the pixmaps
and scrolled regions. After the patch, X runs fine. This sort of corruption
behavior would be disasterous on something like a PCI disk controller.

So I ask that anybody currently maintaining a device which uses the DBDMA
engine to check your code and see if it deals with the DEAD status properly.

Thanks folks!

____________________________________________________________________
 Michael Zucca - mrz5149@acm.org - http://www.mdc.net/~mrz5149/
 "I will choose a path that's clear. I will choose Freewill. "
  --Rush, Freewill
____________________________________________________________________


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: [dmasound] Power Computing DEAD dbdma fix (finally)

[Peter Handel]

Thank you!  This does the job.  Any chance of getting this into the -benh
or Linus kernel?

Just to let you know how much I had to run my "killall xmms; rmmod
dmasound; insmod dmasound" script before your patch:

Aug 30 01:46:51 like kernel: dmasound_pmac: tx-irq: xfer died - patched it up...
Aug 30 01:50:15 like kernel: dmasound_pmac: tx-irq: xfer died - patched it up...
Aug 30 01:51:19 like last message repeated 4 times
Aug 30 01:52:29 like last message repeated 2 times
Aug 30 02:05:01 like last message repeated 2 times
Aug 30 02:06:02 like last message repeated 52 times


Thanks,
Peter


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/