[Gems-users] Alternate control by simics and opal

[sparsh mittal ISU <sparsh@xxxxxxxxxxx>]

Hello All
Regarding the post copied below, I would like to implement transferring the control between opal and simics, so that alternatively opal and simics drive the ruby.

I would like to ask, whether it is currently possible or has any user implemented it. If not, would the authors give me some hint, how should I go about it and which files need to be modified? I would be grateful for any help.
--
Thanks and Regards
Sparsh Mittal
Electrical and Computer Engineering
Iowa State University


===========================Previous Post=======================
Daniele,

  This is a current limitation of Opal.  Once Opal encounters a Simics
magic break instruction, control is returned to Simics, and there is currently no way
of continuing simulation with Opal.

  What I do to get around the problem is the following:  Run with Simics
alone to the magic break (assuming you don't want to have timing info up
to the breakpoint), and once the magic breakpoint is reached load Opal and
Ruby, and continue the rest of the simulation.
    If you need Ruby caches warmed up, you can run with Ruby
alone first until the breakpoint, then load Opal and continue the
remainder of the simulation.

   Luke
On Thu, 21 Sep 2006, Daniele Bordes wrote:

> Hi all. I am running simulations with Simics over a Solaris-Sparc
> simulated machine. I have the following problems:
>
> if I load ruby and opal modules into simics and I run simulation using
> opal command opal0.sim-step N (where N is an integer number of
> istructions to execute), if the simulation "encounters" a
> MAGIC_BREAKPOINT (that is, over Sparc, the instruction sethi 0x40000,
> %g0), simulation is correctly interrupted, but it seems impossible to
> continue it again (using both "continue" or opal0.sim-step). I tried
> to unload opal, but it seems impossible. I have read on the mailing
> list that another user had the same problem.
> Does anyone know what's wrong and how to solve the problem?