Re: pci_dio_32hs read problem followup

hello again,

I found the problem relating to the 'Broken pipe' during
reading from /dev/comedi0 with the ni_pcidio driver.
Disabling the USE_DMA flag (so effectively during DMA off)
and recompiling the driver allows comedi_test and all the
demo/ software to function as intended without the
consistent 'Broken pipe' errors coming from read().

So, my hunch is that my BIOS may be incorrectly configured
to use DMA or there is some problem with the ni_pcidio
driver and its use of DMA.

Does anyone have this driver working with DMA enabled?
(note, only the most recent release of comedi-0.7.64 has DMA
capabilities in this driver).

cheers,
Brent



David Schleef wrote:
> 
> On Fri, May 31, 2002 at 04:12:25PM -0400, brent ledvina wrote:
> > Here is the comedi_test output (notice the presence of many
> > Broken Pipes when attemting a read).  Any ideas as to the
> > cause?
> 
> It is probably starting correctly, but it fills the buffer in
> about 8192*50ns = 400 us and causes the "Broken Pipe" errors.
> 
> The answer lies somewhere between 100 samples/s and 20 MS/s.
> 
> dave...

-- 

--------------------------------------------------------------
Brent M. Ledvina
ECE - Space Plasma Physics

Cornell University
301 Rhodes Hall
Ithaca, NY 14853
v 607.255.4641
f 607.255.6236
--------------------------------------------------------------

Received on 2002-06-06Z15:44:37