- From: David Carr <dc_at_dcarr.org>
- Date: Mon, 29 Mar 2004 10:47:03 -0600
Hi all, Due to a disk failure I've had to rebuild my capture machine. I installed comedi 0.7.68, comedilib 0.7.21 and rtai-3.0r3. I'm using a gentoo machine based on a vanilla kernel.org 2.4.25 kernel. The kernel I'm currently using is not rtai patched but I experienced the same issues with a rtai kernel. My aquisition card is a National Instruments at-mio-16e-2. Analog reads of ch0 work correctly using the 'inp' and 'insn' comedi demos. If I use the command demo, 'cmd' with a number of reads less than 1024 (ie: 1023) everything works too. However if the number of reads is 2048, the value '2048' will be returned 1024 times followed by the correct data for the last 1024 times. Longer reads repeat this pattern with '2048' returned in multiples of 1024 and with the last 1024 values being the correct data. For example, a read of length 5000 returns '2048' 4096 times followed by correct samples 904 times. The only deviation from this pattern is with reads of length between 1025 and 2048. A read of length 1025 returns correct data for the first 1024 values followed by '2048' 1024 times. Note: The buffer on the card is 2048 bytes deep and the data appearing in groups of 1024 would be consistent with a half-full buffer read. I had this setup working with [not quite sure of the versions] comedi 0.7.66, comedilib 0.7.20, kernel 2.4.23 and rtai 24.? I thought that I'd upgrade to the new versions since I had to rebuild the system. What am I doing incorrectly? Thanks for any ideas, David Carr Rice University
Received on 2004-03-29Z15:47:03