Experimental Physics and
Industrial Control System

Gabriel Fedel via Core-talk <core-talk at aps.anl.gov> · Thu, 8 Apr 2021 09:21:21 +0200

Hi Andrew,

On 4/7/21 6:24 PM, Johnson, Andrew N. wrote:

Hi Gabriel,

On Apr 7, 2021, at 2:54 AM, Gabriel Fedel wrote:

After some tests I noticed that this change is causing a different behavior on one of my modules. Reading the discussion and the commit message I understood that this modification shouldn't change the base behavior but I think in my case it is.

Is there some kind of configuration or test that I can do to force to avoid the inheritance?

Can you give us a bit more detail about the change that you’re seeing and how you have narrowed it down to this modification to the Priority Inheritance change? What epicsThread objects are you using, what Linux are you running it on, are you getting a deadlock, etc.

Sure! The issue I'm facing is a low level one and as far I could debug

my IOC is kind of "loosing" one interruption from firmware. It is

something a bit difficult to debug, but it looks to me that the threads

could be having a different priority than before.

I could check that the issue is related to this change reverting the

related commits and compiling all my modules with this patched version.

Without these commits my IOC work as usual, with them I got the problem.

I've tested my IOC on 2 different Linux (CentOS 7 and a Yocto one used

by ESS), both of them with the RT patch.

I'm not sure if I understood the question about the epicsThread, but the

ones are created by one of my submodules have epicsThreadPriorityMedium.

Thanks,

- Andrew

Thank you

--
Gabriel Fedel

Control System Engineer
Integrated Control System Division
The European Spallation Souce
Odarslövsvägen 113
224 84 Lund

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System