Simple 4 lines hang code using Ravenscar. Its this a Gnat bug?

Simple 4 lines hang code using Ravenscar. Its this a Gnat bug?

[Daniel]

Hello,
I found something strange using Ravenscar with Gnat. The code hangs without any explanation when i try to declare a task type...¡even if i dont use any variable of this type!!. If i quit the Ravenscar pragma it works well. This is the code:

--MAIN PROCEDURE
with tmr;
with ada.Text_IO; use ada.Text_IO;
procedure main is
begin
   put_line("Let's hang..");
end main;

--- TMR.ADS FILE
pragma Profile (Ravenscar);
package Tmr is
   task type K_Timer is
   end K_Timer;
end Tmr;

-- TMR.ADB FILE
package body Tmr is
   task body K_Timer  is
   begin
     null;
   end K_Timer;
end Tmr;


Im using GNAT GPL 2017 for windows. 
Thank you and best regards.

Re: Simple 4 lines hang code using Ravenscar. Its this a Gnat bug?

[Niklas Holsti]

On 19-02-22 10:56 , Daniel wrote:
>
> Hello,
> I found something strange using Ravenscar with Gnat. The code hangs without any explanation when i try to declare a task type...¡even if i dont use any variable of this type!!. If i quit the Ravenscar pragma it works well. This is the code:
>
> --MAIN PROCEDURE
> with tmr;
> with ada.Text_IO; use ada.Text_IO;
> procedure main is
> begin
>    put_line("Let's hang..");
> end main;
>
> --- TMR.ADS FILE
> pragma Profile (Ravenscar);
> package Tmr is
>    task type K_Timer is
>    end K_Timer;
> end Tmr;
>
> -- TMR.ADB FILE
> package body Tmr is
>    task body K_Timer  is
>    begin
>      null;
>    end K_Timer;
> end Tmr;
>
>
> Im using GNAT GPL 2017 for windows.
> Thank you and best regards.
>

In a Ravenscar program, no task should terminate, and this includes the 
environment task. However, your "main" subprogram, which is where the 
environment task ends up after elaboration, does terminate. Therefore no 
Ravenscar program is expected to terminate at all!

My guess is that without the task-type declaration, the tasking part of 
the Ravenscar run-time is omitted from the program, and the environment 
task can terminate and this terminates the program. With the task-type 
declaration, the Ravenscar tasking run-time is included, and termination 
of the environment task becomes a no-no. Or at least it no longer 
terminates the program, which therefore seems to "hang".

This is just a guess, of course.

You could try to put an eternal loop in the "main", say

    loop
       Put_Line ("Idling...");
       delay until Ada.Real_Time.Clock + Ada.Real_Time.Seconds (1);
    end loop;

HTH

-- 
Niklas Holsti
Tidorum Ltd
niklas holsti tidorum fi
       .      @       .

Re: Simple 4 lines hang code using Ravenscar. Its this a Gnat bug?

[AdaMagica]

pragma Profile is a configuration pragma, and as such it must be defined (for Gnat) in a special configuration unit.
Niklas has answered the prolem with Ravenscar.

Re: Simple 4 lines hang code using Ravenscar. Its this a Gnat bug?

[Daniel]

El viernes, 22 de febrero de 2019, 11:01:51 (UTC+1), Niklas Holsti  escribió:
> On 19-02-22 10:56 , Daniel wrote:
> >
> > Hello,
> > I found something strange using Ravenscar with Gnat. The code hangs without any explanation when i try to declare a task type...¡even if i dont use any variable of this type!!. If i quit the Ravenscar pragma it works well. This is the code:
> >
> > --MAIN PROCEDURE
> > with tmr;
> > with ada.Text_IO; use ada.Text_IO;
> > procedure main is
> > begin
> >    put_line("Let's hang..");
> > end main;
> >
> > --- TMR.ADS FILE
> > pragma Profile (Ravenscar);
> > package Tmr is
> >    task type K_Timer is
> >    end K_Timer;
> > end Tmr;
> >
> > -- TMR.ADB FILE
> > package body Tmr is
> >    task body K_Timer  is
> >    begin
> >      null;
> >    end K_Timer;
> > end Tmr;
> >
> >
> > Im using GNAT GPL 2017 for windows.
> > Thank you and best regards.
> >
> 
> In a Ravenscar program, no task should terminate, and this includes the 
> environment task. However, your "main" subprogram, which is where the 
> environment task ends up after elaboration, does terminate. Therefore no 
> Ravenscar program is expected to terminate at all!
> 
> My guess is that without the task-type declaration, the tasking part of 
> the Ravenscar run-time is omitted from the program, and the environment 
> task can terminate and this terminates the program. With the task-type 
> declaration, the Ravenscar tasking run-time is included, and termination 
> of the environment task becomes a no-no. Or at least it no longer 
> terminates the program, which therefore seems to "hang".
> 
> This is just a guess, of course.
> 
> You could try to put an eternal loop in the "main", say
> 
>     loop
>        Put_Line ("Idling...");
>        delay until Ada.Real_Time.Clock + Ada.Real_Time.Seconds (1);
>     end loop;
> 
> HTH
> 
> -- 
> Niklas Holsti
> Tidorum Ltd
> niklas holsti tidorum fi
>        .      @       .

It really make sense. Thank you.

Re: Simple 4 lines hang code using Ravenscar. Its this a Gnat bug?

[Simon Wright]

AdaMagica <christ-usch.grein@t-online.de> writes:

> pragma Profile is a configuration pragma, and as such it must be
> defined (for Gnat) in a special configuration unit.

I'm not sure what the rules are, and it's certainly true that putting
the configuration pragmas in a special unit (typically, and with
gnatmake automatically effective if found, gnat.adc).

But, if they're for use in a [GNAT?] configurable runtime, it's best to
put them in system.ads so there's no chance of confusion when compiling
against the runtime.

Re: Simple 4 lines hang code using Ravenscar. Its this a Gnat bug?

[Simon Wright]

AdaMagica <christ-usch.grein@t-online.de> writes:

> pragma Profile is a configuration pragma, and as such it must be
> defined (for Gnat) in a special configuration unit.

I'm not sure what the rules are, and it's certainly true that putting
the configuration pragmas in a special unit (typically, and with
gnatmake automatically effective if found, gnat.adc).

But, if they're for use in a [GNAT?] configurable runtime, it's best to
put them in system.ads so there's no chance of confusion when compiling
against the runtime.