Re: GPIB driver

[timmer@luke.atdiv.lanl.gov (Carl Timmer)]

To those of you doing GPIB:

There is a problem that surfaced recently concerning the NI 1014D
gpib board. When doing DMA data reads (i.e. reading waveforms
from a Tektronix digitzer), bus errors were mysteriously generated.

It turns out that some changes to the "drvGpib.c" code would be advisable.
In the "niPhysIo" routine, there are two lines,

  b->ch0.ccr = D_SRT;
  b->imr2 = w_imr2;	/* this must be done last */

between which there should be a time delay. Apparently, the gpib board is
in occasional violation of the VME standard and the attempt to write to
interrupt mask register 2 (I believe) fails. I have a fix which also allows
one to fix & debug the situation. If interested, send me email and I'll send a
revised copy of "drvGpib.c" .

This correction will be included into the next release of EPICS.

Carl Timmer