Re: [Gems-users] Ruby initialization fails in simics 3.0.18

[Dan Gibson <degibson@xxxxxxxx>]

I spent a lot of time figuring out this issue with Simics 3.0.5. I 
finally tracked down the segmentation fault to something going on within 
the standard C++ string library, which suggested memory corruption.
The root of the issue is that Simics is linked against an older version 
of glibc, which handles memory management a little differently than more 
modern versions. We made a simple module to verify the issue.
To figure out if the version incompatibility is indeed the issue, please 
do the following:
Open $GEMS/ruby/init.C and place the following lines at the start of the 
init_simulator() function:
cout << "Hello World" << endl << flush;
string desc = string("Hello World, ")+string("01")+string(" 
")+string("02")+string(" ")+string("03");
cout << desc << endl << flush;

If you see Hello World followed by a segmentation fault, we're dealing 
with the same problem.
If you see the whole string, we might be dealing with a separate 
problem. I say *might* because it is likely that changes from 3.0.5 -> 
3.0.18 have fixed the above problem and introduced new ones -- they 
dynamic interaction of memory allocations/deallocations are enormously 
difficult to debug.
IF a glibc compatibility issue is at the root of the problem, there are 
essentially three options:
1) Use an old version of gcc/glibc to compile ruby -- the same version 
used by Simics would be ideal.
2) Ask Virtutech to do one of the following:
A) Provide a copy of the Simics 3.0.11 tarball (known to work with Ruby) 
(fat chance)
B) Provide a compiled version of Simics that used a modern gcc/glibc. 
(fat chance)
3) Use Simics 2.2.19 instead of 3.0.X.

Regards,
Dan

Mishali Naik wrote:
> I am using gcc 3.4.4. thanks for the prompt reply. I have spent quite
> sometime trying to figure out...
>
> On 9/11/06, Dan Gibson <degibson@xxxxxxxx> wrote:
>   
> > This sounds like an issue I ran into some time ago relating to glibc.
> > Essentially, there was a version incompatibility between my compiler the
> > compiler used to compile Simics.
> >
> > What version of gcc are you using?
> >
> > Regards,
> > Dan Gibson
> >
> > Mishali Naik wrote:
> >     
> > > Hi,
> > >         I am trying to initialize the ruby module with ruby0.init in
> > > simics 3.0.18. But the initialization fails because of a segementation
> > > fault:
> > >
> > > Turning I-STC off and flushing old data
> > > Turning D-STC off and flushing old data
> > > successful installation of the ruby timing model.
> > > Ruby Timing Mode
> > > Creating event queue...
> > > Creating system ...
> > >     Processors : 8
> > > Segmentation fault (SIGSEGV) in main thread.
> > > The simulation state has been corrupted. Simulation cannot continue.
> > >
> > >            I traced the calls and the assertion in
> > > pc_step_t SIM_step_count(conf_object_t *p) { assert(0); return 0; };
> > > //fails for the processor 1. This function is being called by
> > > integer_t SIMICS_get_insn_count(int cpuNumber)
> > > {
> > >   processor_t* cpu = SIM_proc_no_2_ptr(cpuNumber);
> > >    return SIM_step_count((conf_object_t*) cpu);
> > >
> > > }
> > >
> > > Processor 0 gets initialized fine.
> > >
> > >                             I appreciate all your help.
> > >
> > > thanks,
> > > Mishali
> > > _______________________________________________
> > > 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!
> >
> > _______________________________________________
> > 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!