If I remember correctly, it is possible to manually edit your checkpoint
files, find the line that defines the physical memory object (usually
called phys_mem0, occasionally just phys_mem), and remove "ruby0" from
the timing model attribute.
You should also remove "ruby0" from the object list... just grep for
"ruby" in the check file. Regrettably, Simics (pre 3.0.X) doesn't
support module unloading.
Alternatively, you can modify ruby to call SIMICS_remove_timing_model
before writing the configuration, though ruby will probably still end up
as part of the checkpoint.
Regards,
Dan
mehmetderin.harmanci@xxxxxxx wrote:
Hello,
I have a similar problem as Kushal, however it is a slightly different
case. I could not solve how to solve it. The problem is as follows:
Well, what I would like to do is to store a checkpoint during my simulation
where Ruby is also running.
That's, I start simics with a checkpoint and load Ruby, set the parameters
I want and start the simulation. Then I stop the simulation and write the
simics configuration to a conf file. When I restart simics from this
configuration simics gives the "Physical Memory Object not found" error.
I guess simics tries to load ruby module before loading other parts of
the checkpoint which are needed by ruby. Is there a turn around for this?
Or should I run all my ruby simulations from the very beginning all the
time and can not get intermediate checkpoints?
Sorry for interrupting,
Derin Harmanci
Quoting Dan Gibson <degibson@xxxxxxxx>:
Great job diagnosing the first error. I'll probably add that to the FAQ
sometime soon.
Your new error is caused by not reading a configuration before loading
Ruby. At load-time, Ruby searches for a physical memory object to
attach itself to-- there is no physical memory object before a
checkpoint is loaded.
Regards,
Dan
Datta, Kushal wrote:
Hi Dan,
Thanks for your advice. It worked for gcc-4.0.3. Here's the deal. For every
makefile, I had to change the LDFLAGS and MODULE_LDFLAGS variables to point
to the non-native lib folder. Moreover, the SHELL variable in the
common/Makefile.common was set to /bin/sh whereas I was using /bin/bash. At
the end, I checked ruby.so, first using ldd and then objdump -T. All the
proper files were linked and ruby.so had just one entry for ruby_session_set
whereas previosuly it had two entries.
When I first start simics from the MOSI_SMP_bcast directory, and do a
"load-module ruby" it gives me the error as:
/***************************************************************************\
> Physical Memory object cannot be found. If you are NOT compiling Ruby and
<
> you see this message, something is wrong.
<
> This message is part of the normal compilation process.
<
\***************************************************************************/
Although, the gems-1.3.1 ruby.C file has that patch, still it is throwing
this error for the first time. The second time it loads ruby.
Now if I try to run the command "ruby0.setparam g_NUM_PROCESSOR 1" in simics,
it is giving me an error message as:
"attribute param in object ruby0: not found"
In fact, although it shows all the attributes of ruby0 when I press TAB, but
it cannot find any of them and throws the above error message for all of
them.
Thanks and regards,
Kushal.
_______________________________________________
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!
|
![[Computer Systems Lab]](https://www-auth.cs.wisc.edu/pics/csl_logo.gif){kind=logo}