[sv-ec] RE: [sv-ac] Re: sv-ac: example error in "9.7 Fine-grain process control"?

I believe that the only things that need to change is the array (job) declaration to a dynamic array (and the corresponding indexing changes), which would make the example strictly correct:

task do_n_way( int N );
process job[] = new[N];

for ( int j = 1; j <= N; j++ )
fork
automatic int k = j;
begin job[k] = process::self(); ... ; end
join_none

for( int j = 0; j < N; j++ ) // wait for all processes to start
wait( job[j] != null );

job[0].await(); // wait for first process to finish

for ( int k = 0; k < N; k++ ) begin
if ( job[k].status != process::FINISHED )
job[k].kill();
end
endtask

I don't agree with Ben's proposed change to add the fork in the second loop. The sequential check is indeed intended - to wait for all processes to start. Ben's proposed change would break that intent and would wait for no process: the wait processes would simply execute concurrently but with no synchronization.

                Arturo

From: owner-sv-ec@eda.org [mailto:owner-sv-ec@eda.org] On Behalf Of Bresticker, Shalom
Sent: Wednesday, February 22, 2012 7:05 AM
To: sv-ac@eda-stds.org; sv-ec@eda.org
Subject: [sv-ec] RE: [sv-ac] Re: sv-ac: example error in "9.7 Fine-grain process control"?

Hi,

This was previously reported in Mantis 2559.

It has not been fixed.

It could be a ballot comment.

Shalom

From: owner-sv-ac@eda.org<mailto:owner-sv-ac@eda.org> [mailto:owner-sv-ac@eda.org]<mailto:[mailto:owner-sv-ac@eda.org]> On Behalf Of ben cohen
Sent: Sunday, January 01, 2012 5:58 PM
To: sv-ac@eda-stds.org<mailto:sv-ac@eda-stds.org>; stuart@sutherland-hdl.com<mailto:stuart@sutherland-hdl.com>
Cc: Korchemny, Dmitry
Subject: [sv-ac] Re: sv-ac: example error in "9.7 Fine-grain process control"?

After more thoughts, the model work without the fork join, even though it might be a better style. However, the first reported issue is still there:
task do_n_way( int N ); // line 98
process job[1:N]; // (99): Range must be bounded by constant expressions.
Ben Cohen

On Sat, Dec 31, 2011 at 8:47 PM, ben cohen <hdlcohen@gmail.com<mailto:hdlcohen@gmail.com>> wrote:
In code below (copied mostly from LRM), I think that this statement is incorrect:
 for( int j = 1; j <= N; j++ ) // wait for all processes to start
   wait( job[j] != null );

I think that this should be
 for( int j = 1; j <= N; j++ ) // wait for all processes to start
       fork
   wait( job[j] != null );
       join_none
Without the fork, these wait(job...) are sequential.
From 9.4.3 Level-sensitive event control
The wait statement shall evaluate a condition; and, if it is not true (as defined in 12.4), the procedural statements
following the wait statement shall remain blocked until that condition becomes true before continuing.

Attached is a modified model that I constructed to test the LRM example using a simulator. The simulator yielded:
run 100ns
# @t= 0, waiting for jobs[ 1] to start
# @t= 0, waiting for jobs[ 2] to start
# @t= 0, waiting for jobs[ 3] to start
# @t= 2, in task, k= 1
# @t= 3, in task also, k= 1
# @t= 3, killing job 2
# @t= 3, killing job 3

I suggest that the example be modified to reflect code that compiles and reflects the intent of the example.
The way it is now, I believe that the LRM model is in error in 2 respects.
Ben Cohen
systemverilog.us<http://systemverilog.us>

On Sat, Dec 31, 2011 at 5:34 PM, ben cohen <hdlcohen@gmail.com<mailto:hdlcohen@gmail.com>> wrote:
Below is code copied from LRM, draft 4 (slightly modified to comment out the "...;" and add a call to the task.
A simulator cage me this error code upon compilation:
(99): Range must be bounded by constant expressions
Is LRM is error here?
Thanks,
Ben

----
module mprocess;  // line 96
task do_n_way( int N ); // line 98
process job[1:N];   // (99): Range must be bounded by constant expressions.
for ( int j = 1; j <= N; j++ )
fork
automatic int k = j;
begin job[k] = process::self();
$display(" in task, k=%d", k);
// ... ;
end
join_none
for( int j = 1; j <= N; j++ ) // wait for all processes to start
wait( job[j] != null );
job[1].await(); // wait for first process to finish
for ( int k = 1; k <= N; k++ ) begin
if ( job[k].status != process::FINISHED )
job[k].kill();
end
endtask
initial begin : init
do_n_way(3);
end  :init
endmodule :mprocess
--
This message has been scanned for viruses and
dangerous content by MailScanner<http://www.mailscanner.info/>, and is
believed to be clean.
---------------------------------------------------------------------
Intel Israel (74) Limited
This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.
--
This message has been scanned for viruses and
dangerous content by MailScanner<http://www.mailscanner.info/>, and is
believed to be clean.
-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.