Re: [Gems-users] Opal wont catch magic breakpoint

[Xuan Qi <xqi@xxxxxxxxxxx>]

Thanks for the comments. I make the magic breakpoint work but I have a 
follow-up question.
I check the system.C and found this piece of code:

system_breakpoint( void *data, conf_object_t *cpu, integer_t parameter )
{
  if ( parameter != 4UL << 16 ) {
    // ignore all transaction completion calls
    /* sprintf( system_t::inst->m_sim_message_buffer,
             "other breakpoint (ignoring) 0x%llx", parameter );
    */
    //cout << "system_breakpoint called" << endl;
    // MAGIC breakpoints get intercepted here.  Opal currently does not do 
anything special here
    return;
  }

  //currently IGNORE the system_breakpoint (parameter = 0x4000)...
  #if 0
  ERROR_OUT("system_t::system_breakpoint REACHED param[ 0x%x ]\n", 
parameter);
  sprintf( system_t::inst->m_sim_message_buffer,
           "magic breakpoint reached" );
  HALT_SIMULATION;
  return;
  #endif
}

I changed #if 0 ==> #if 1 then seems the system_breakpoint is enabled and
my toy program will stop at the right place. Just curious why 
system_breakpoint is ignored in the original code? Thanks...
Best regards,
Xuan Qi
PHD Student
Department of Computer Science
The University of Texas at San Antonio


On Mon, 10 Mar 2008, Dan Gibson wrote:

> While the loop appears simple, Luke could be right in his observation
> that 100000 instructions may not be enough. In particular, I am unsure of
> the implementation of fabs() -- if it calls a dynamically-linked library
> then it could take quite some time! Try calling fabs() before the first
> magic call, just to be sure.
>
> Regards,
> Dan
>
> Luke Yen wrote:
>
>  The opal0.sim-step command takes as argument number of processor 
> instructions to execute, so 100000 might be insufficient for the 
> processor to reach the second breakpoint.
>     Also check whether the breakpoint is caught inside Opal's magic 
> instruction callback handler, in opal/system/system.C.  Grep for 
> system_breakpoint() and insert some print statements.
>     Luke
>
> On Mon, 10 Mar 2008, Xuan Qi wrote:
>
>
>
>  Hi everyone,
>
> I am testing the magic breakpoint on my target machine. I use the a simple
> program which is provided at $GEMS/opal/benchmark/float.C. The main part
> of it is:
>
> ......
>  /* -- 4 == Do_Breakpoint */
>  RUBY_MAGIC_CALL( 4 );
>
>  for (int i = 0; i < numTrials; i++) {
>
>    result_s = ((float32) fabs( pos1_s )) * value_s ;
>    result_s = ((float32) fabs( neg1_s )) * value_s ;
>
>    result_d = (fabs( pos1_d )) * value_d ;
>    result_d = (fabs( neg1_d )) * value_d ;
>
>  }
>
>  /* -- 4 == Do_Breakpoint */
>  RUBY_MAGIC_CALL( 4 );
> ......
>
> And I use the following script to test the magic breakpoints:
>
> read-configuration sarek-2core.conf
> cpu-switch-time 1
> instruction-fetch-mode instruction-fetch-trace
> istc-disable
> dstc-disable
> magic-break-enable
> load-module ruby
> ruby0.setparam g_NUM_PROCESSORS 2
> ruby0.setparam g_MEMORY_SIZE_BYTES 536870912
> ruby0.setparam g_PROCS_PER_CHIP 2
> ruby0.setparam g_NUM_MEMORIES 1
> ruby0.setparam NUMBER_OF_VIRTUAL_NETWORKS 6
> ruby0.init
> @conf.con0.input = "cd ALP/MPGenc/verify\n";
> @conf.con0.input = "./float 5\n";
> c
> echo "first breakpoint"
> load-module opal
> opal0.init
> opal0.sim-start "output_2core.opal"
> opal0.sim-step 100000
>
> Simulation will stop at first magic breakpoint and load opal module. But
> when opal drives simulation, it won't catch the second magic breakpoint.
> Am I missing something? Thank you...
>
> Best regards,
> Xuan Qi
>
>
> _______________________________________________
> Gems-users mailing list
> Gems-users@xxxxxxxxxxx
> https://lists.cs.wisc.edu/mailman/listinfo/gems-users
> Use Google to search the GEMS Users mailing list by adding "site:https://lists.cs.wisc.edu/archive/gems-users/"; to your search.
>
>
>
>  _______________________________________________
> Gems-users mailing list
> Gems-users@xxxxxxxxxxx
> https://lists.cs.wisc.edu/mailman/listinfo/gems-users
> Use Google to search the GEMS Users mailing list by adding "site:https://lists.cs.wisc.edu/archive/gems-users/"; to your search.
>
>
>
>
>
>  -- 
> http://www.cs.wisc.edu/~gibson [esc]:wq!