Experimental Physics and
Industrial Control System

Marty Kraimer <[email protected]> · Wed, 21 Feb 1996 12:43:41 -0600

Andy Foster wrote:
> 
> I have some questions/comments on EPICS "Access Security":
> 
> (1) I wanted to see the effect of stopping a user from accessing
> data values with "dm", by using a suitably set-up access security file.
> It did work. However, I was somewhat surprised by the message returned in
> the "dm" text box which was "No data available". It was only when I looked
> at messages in my command tool that I realised that access was being denied.
> I don't see at present, how the user, just from looking at the "dm" screen could
> tell that access was being denied? Couldn't "dm" be made clever enough to distinguish
> the cases of "No data available" and "Insufficient Access"? At 14,000 ft
> misleading messages will lead to even greater confusion when a system is
> being debugged!
> 

What medm does is display a sepcial icon over something with no read access.
Similarly if you move into a controller for which write access is denied
a special icon is displayed. It is quite intutitive.

I will let Deb or Bob explain what the plans are for dm.

> (2) In another scenario, I was making a database in IOC1 send a command to
> a database in IOC2. By sending a command, I mean writing to a field of a record
> in IOC2 which had ASL = 0 and PP = YES in the ascii file.
> I found that when the command
> was blocked by the access security mechanism, the following appeared on the IOC1
> shell (sender of the command):
> 
> ************************************************************************
> task: 0Xe7d640 dbCaOutput
> dbCa_unknownECA ERROR: queue_ca_put() unrecognizable status returned 376
> 
> Bus Error
> Program Counter: 0x00f0bc72
> Status Register: 0x3000
> Access Address : 0xf8609ffc
> Special Status : 0x0485
> Task: 0xe4f338 "process_asynch_events_task"
> **************************************************************************
> 
> When I saw "Bus Error" I was rather shocked. Should blocking out access really
> cause such an error to occur in the sending IOC or is this indicative of a
> more serious error?

I am also shocked!! Sure looks like a bug. Bob Dalesio, as part of adding support
to dynamically change link fields, is redoing the dbCa support. This should
be added as something to check.

Marty Kraimer

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System