Experimental Physics and
Industrial Control System

Andrew Johnson <[email protected]> · Mon, 9 Oct 1995 12:12:41 +0100 (BST)

On Fri, 06 Oct 1995 17:05:40 EDT Bonnie Montjar <[email protected]> wrote:
>I am using the XYCOM 240 board configured for 32 inputs (ports 0,1,2,3) and
>32 outputs (the rest).  I've loaded the card into test crate with an
>ioc (162), and set up my database to have it read data bits on the first
>two ports.  Well, when I checked it by poking the hardware, where I
>thought it should drive the bit low, I got no response (ports 0&1), 
>but when I tried the same thing on what, according to the XYCOM 240
>manual, are ports 2&3, I got little red lights all over the screen!
>Anybody else run into this? 
>
>Here's the wierd part - my hardware addresses say the database is looking
>at the data bits in ports 0 & 1, but....  Have I been sitting in front
>of a radiation-producing terminal too long?  Am I missing something
>really obvious?

Bonnie,

What version of EPICS are you running?  
How are you doing the conversion from Signal Number to port,bit?  

When we started using the xy240 at the RGO about a year ago we found the driver had 
bugs in, relating to addressing and getting port data confused.  Unfortunately the 
driver fix that we produced got "refixed" before it was installed, so the 3.12.1 
release driver still has buggs in it (N.B. for Gemini developers, the Gemini 
releases work OK).  Our fix *is* in the Beta for 3.12.2, and I will attach the 
patch file for it (from the 3.12.1 version) at the end of this message.

The other vague possibility is that the driver fix we produced may have changed the 
way the card bits are addressed from the way you're using it -- I hope not, but I'm 
not sure what documentation (if any) you're working from.

This is how the 3.12.1 driver maps signal numbers to ports:
	The I/O bits appear in the memory space as 4 adjacent bytes, so are read 
and written by the driver as a single 32-bit number.  Port 0 is the MSByte of this 
32-bit value & port 3 the LSByte, therefore P3B0 maps to S0 and P0B7 maps to S31.

If this is different to the way you're used to, please let me know because we have 
a problem which needs to be resolved - the RGO and Gemini databases use this 
mapping.  The original driver didn't actually implement the other mapping properly 
anyway, so no one else seems to have been using it.

- Andrew

  O o ._    Andrew Johnson, Royal Greenwich Observatory
     __\\_	Madingley Road, Cambridge, CB3 0EZ
  __|ooooo|__	Tel +44 (0)1223 374823  Fax 374700
~~\_________/~~	WWW http://www.ast.cam.ac.uk/~anj/

--- drvXy240.c.orig     Thu May 25 11:48:30 1995
+++ drvXy240.c  Thu May 25 11:48:30 1995
@@ -338,13 +338,13 @@
        /* use structure to handle high and low short swap */
        /* get current output */
 
-       work = (dio[card].dptr->port6_7 << 16)
-                        + dio[card].dptr->port4_5;
+       work = (dio[card].dptr->port4_5 << 16)
+                        + dio[card].dptr->port6_7;
 
        work = (work & ~mask) | (val & mask);
 
-       dio[card].dptr->port6_7 = (unsigned short)(work >> 16);
-       dio[card].dptr->port4_5 = (unsigned short)work;
+       dio[card].dptr->port4_5 = (unsigned short)(work >> 16);
+       dio[card].dptr->port6_7 = (unsigned short)work;
 
        return OK;
  }




Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System