Experimental Physics and
Industrial Control System

Jane Richards <[email protected]> · Fri, 14 Feb 2014 15:55:58 -0800

Hi there.

We use a linux VME kernel driver and API from GE-Fanuc. We have created

a "software shim" that recreates the VxWorks API in terms of the Linux

driver. This was developed using EPICS release 3.14.8.2 and has worked

well up to and including 3.14.11. However release 3.14.12 introduced a

"devLib

cleanup" which changed the API and broke our code. We now want to

update our code to a sustainable model.

How should we proceed?

BTW We require devlib2 for building support for the MRF event

generator/receiver which will also run on Linux. How does devlib V1

relate to the additional functions defined in devlib2?

Thanks,
Jane

On 02/14/2014 02:34 PM, Rod Nussbaumer wrote:

On 02/14/2014 01:49 PM, Jane Richards wrote:

Hi Rod,

For your input...
Jane

On 02/14/2014 12:45 PM, graham waters wrote:

Hi Jane
Before I send anythingtech-talk I should run it by you first.

Graham

Subject: linux vme support

The below statement is similar to those which I consistently find

troubling; the blurring of functionality at the kernel module level,

and the EPICS level. The wording needs to make it clear what level is

referred where it says 'and mimic vxWorks functions calls'. I assume

we mean that the vxWorks mimicry is done in some level of our local

EPICS code. This should be made clear, to avoid any confusion by

readers. Is there some level of code that we add, replace, or modify

that is part of EPICS base &/or devLibX? Specify.

We use a linux vme kernel driver and API from GE-Fanuc and mimic

vxWorks functions calls in order to use the osi functionality of

devlib.

This was developed using EPICS release 3.14.8.2 and has worked well up
to and including 3.14.11. However release 3.14.12 introduced a "devLib
cleanup" which removed a structure that I rely on - "devLibVirtualOS". I
realize that my Linux support was not registered with the EPICS base
developers - mea culpa. I now want to update my code to a sustainable
model.

My questions are
1) What was the rational behind the changes and how should I proceeed.
2) How does devlib V1 relate to the additional functions defined in
devlib2-2.4 which we downloaded from sourceforge

We require devlib2 for building support for the MRF event
generator/receiver which I will also be porting to Linux.

--
Jane Richards
Controls Group Leader, TRIUMF
4004 Wesbrook Mall, Vancouver, BC V6T 2A3
Phone: (604) 222-1047 (ext 7693)
Fax: (604) 222-1074

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System