Experimental Physics and
Industrial Control System

Andrew Johnson <[email protected]> · Mon, 21 Jul 2008 14:34:12 -0500

On Friday 18 July 2008 13:15:03 Mark Rivers wrote:
>
> Where do things stand in EPICS in terms of running IOCs in 64-bit mode?

Not recommended with R3.14.9 although many things do work; the problems are in 
the more obscure areas like numeric precision values when fetching compound 
DBF types.  I have put quite a lot of work recently into trying to round up 
64-bit issues in the IOC code, so I hope that R3.14.10 should be good for 
64-bit IOCs.

Note that you can't just go through and replace every 'long' with 
an 'epicsInt32' in the source code because that would break source-code 
compatibility with earlier versions.  I think what I have should be source 
compatible with most external device support and record types, although some 
changes might be necessary since DBF_LONG fields that used to be defined as a 
long are now an epicsInt32.

> However, a 64-bit IOC to control the Pilatus detector is failing.  It
> seems to be crashing in SNL code.  What parts of EPICS (base 4.14.9,
> asyn, SNL etc) are believed to be 64-bit safe and which are not?

I haven not looked at SNL or Asyn at all for 64-bit compatibility, nor the 
portable CAS in Base.  The Base libCom routines and the client side of CA are 
fine on 64-bit systems in R3.14.9.

- Andrew
-- 
Talk is cheap. Show me the code. -- Linus Torvalds

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System