Re: Aborting requeued entry calls

Re: Aborting requeued entry calls

[Robert A Duff]

Dmitry A.Kazakov <mailbox@dmitry-kazakov.de> writes:

> Hi!
> 
> I have a problem and a question. Consider the following:
> 
> package Test_Abort is
>    protected Thing is
>       entry Foo_1;
>       entry Foo_2;
>    end Thing;
> end Test_Abort;
> ------------------------
> with Ada.Text_IO;  use Ada.Text_IO;
> 
> package body Test_Abort is
>    protected body Thing is
>       entry Foo_1 when True is
>       begin
>          Put_Line ("Foo_1 requeues to Foo_2");
>          requeue Foo_2; -- w/o abort
>       end Foo_1;
>       entry Foo_2 when False is
>       begin
>          null;
>       end Foo_2;
>    end Thing;
> end Test_Abort;
> -------------------------
> with Test_Abort;   use Test_Abort;
> with Ada.Text_IO;  use Ada.Text_IO;
> 
> procedure Test is
> begin
>    select
>       delay 1.0;
>    then abort
>       Thing.Foo_1;
>    end select;
>    Put_Line ("Why is that abortable?");
> end Test;
> --------------------------
> 
> The problem. From what I read about requeue with no abort I would expect 
> that Test should hang forever.

Yes, I think that's correct.

- Bob

Aborting requeued entry calls

[Dmitry A.Kazakov]

Hi!

I have a problem and a question. Consider the following:

package Test_Abort is
   protected Thing is
      entry Foo_1;
      entry Foo_2;
   end Thing;
end Test_Abort;
------------------------
with Ada.Text_IO;  use Ada.Text_IO;

package body Test_Abort is
   protected body Thing is
      entry Foo_1 when True is
      begin
         Put_Line ("Foo_1 requeues to Foo_2");
         requeue Foo_2; -- w/o abort
      end Foo_1;
      entry Foo_2 when False is
      begin
         null;
      end Foo_2;
   end Thing;
end Test_Abort;
-------------------------
with Test_Abort;   use Test_Abort;
with Ada.Text_IO;  use Ada.Text_IO;

procedure Test is
begin
   select
      delay 1.0;
   then abort
      Thing.Foo_1;
   end select;
   Put_Line ("Why is that abortable?");
end Test;
--------------------------

The problem. From what I read about requeue with no abort I would expect 
that Test should hang forever. Yet GNAT under Windows successfully
aborts Thing.Foo_1 (after it was requeued to Foo_2).

Is that OK?

If yes, then the question, if a requeued entry call may be silently removed 
from the wait queue, how Thing could become aware of that? Let Foo_1 
changes the state of Thing and Foo_2 restores it, then if Foo_2 get 
silently aborted Thing becomes corrupted. Any solution?

Thanks.

-- 
Regards,
Dmitry Kazakov
www.dmitry-kazakov.de

Re: Aborting requeued entry calls

[Dmitry A. Kazakov]

On Fri, 2 Aug 2002 15:33:17 GMT, Robert A Duff
<bobduff@shell01.TheWorld.com> wrote:

>Dmitry A.Kazakov <mailbox@dmitry-kazakov.de> writes:
>> 
>> The problem. From what I read about requeue with no abort I would expect 
>> that Test should hang forever.
>
>Yes, I think that's correct.

Thank you very much for the response!

[ It was so distressing when I discovered that three months of design
were probably wasted for nothing that I could not find a rest for two
days! ]

---
Regards,
Dmitry Kazakov
www.dmitry-kazakov.de