Re: Re: Questions related to mcoreutil

[김대영 via Tech-talk <tech-talk@aps.anl.gov>]

 

Dear Ralph Lange

Thank you for your help every time.

As a result of confirmation, Thread is created and executed before mcoreutil is executed as follows.

 

Q1) How can mcoreutil run faster?

   I tried to give sleep before creating a thread.

    And I changed the position of XXX_LIBS += mcoreutils in the make file.

   The generated Thread is generated from Main.cpp, unlike other threads, as shown in the figure below.

 

Thank you and Best Regards

--- Original Message ---
From : "Ralph Lange"<ralph.lange@gmx.de>
To : "김대영"<dykim1114@dawonsys.com>, tech-talk@aps.anl.gov
Date : 2023/02/21 화요일 오후 6:11:32
Subject : Re: Questions related to mcoreutil

Hi Daeyoung,

On Tue, 21 Feb 2023 at 09:13, 김대영 via Tech-talk < tech-talk@aps.anl.gov> wrote:

 

I wrote and uploaded the file of /etc/rtrules as follows.

The results are as follows. However, some of them, such as Thread_ExchangeData and CAC_UDP, were not CPU-assigned.

 

Obviously, some of the rules are applied, which means the rule mechanism is reading the file and generally working. You can check if all rules have been read correctly with the command ' mcoreThreadRulesShow' in the IOC shell.

The rules work through a hook in the mechanism behind ' epicsThreadCreate()', the EPICS Base call to create threads. Any threads that already have been created before MCoreUtils adds itself to the hook will obviously not be affected.

Looking at the 'EPICS ID' values of the un-assigned threads, which - except for those with stack addresses - are all starting with '0x1f...' while the threads with working assignments all start with '0x20...' I think the installation of the rules might be happening too late in the IOC startup.

There should be a line of log printout on the IOC console:

MCoreUtils: Read <n> thread rule(s) from <file>

Do you see that line? When is it printed?

Cheers,
~Ralph