Subject: |
Re: Logical Naming of Hardware Addresses |
From: |
Andrew Johnson <[email protected]> |
Date: |
Thu, 26 May 1994 10:17:21 BST |
On Wed, 25 May 94 12:42:18 -0500 [email protected] wrote:
> Andrew Johnson talked about problems he had configuring reflective memory.
> I have no already existing "correct" solution. However all of us involved
> with ioc development recognize the problem and really want to come up with
> a nice solution.
...
> The solution is to replace hardware links by a reference to a new type of
> record. This record can contain fields just like existing record types. The
> difference is
> 1) The record has no database common
> 2) It is not processed.
>
> For each device type the person providing device support provides an ascii
> record definition for the device. This record can contain all the fields
> needed to configure an instance of the device. For example a vme device record
> would have fields to define all the info in the existing vme link definition.
> The record could also contain any additional fields the developer decides
> are needed.
>
> Database Generation Tools would allow a user to create a record instance for
> each device instance.
>
> The existing link definitions would replace the hardware definitions by a
> definition that allows the link to reference this new type of record, i.e.
> this provides logical hardware links.
This is an interesting idea for interfacing to hardware (I can see it would
allow more flexibility in device configuration), but I'm not sure it solves my
problem. I have to ensure that all IOCs have the same idea as to whereabouts in
the Reflective Memory a particular record lives, and I'm reluctant to rely on
the database designer calculating and inserting "Magic Numbers".
Other than by extending the database tools to add specific support code, the
only way I can see this happening automatically is if every IOC is told about
every RM record (even the ones which this IOC has no other interest in) so the
driver can work the addresses out for itself. That means adding a lot of
redundant "hardware" records to a database, and no longer provides a single
source of RM address information because each database has its own copy of
these.
_ _
/ \ ___ _| |_ _ ___ _ _
| - | | . | ~_| -_| |~| |
|_|_|_|_|___|_| |___|_____|
- Navigate by Date:
- Prev:
Re: Logical Naming of Hardware Addresses watson
- Next:
Reflections on a Reflective Memory Device Interface Bob Dalesio
- Index:
<1994>
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
- Navigate by Thread:
- Prev:
Re: Logical Naming of Hardware Addresses watson
- Next:
Reflections on a Reflective Memory Device Interface Bob Dalesio
- Index:
<1994>
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
|