Hi Ali,

      Thanks for your reply, i changes the following in different way still
it is not working

    /* 116 */ SyscallDesc("gettimeofday", gettimeofdayFunc), //32 bit
    /* 116 */ SyscallDesc("gettimeofday", gettimeofdayFunc<SparcLinux>),
//32 bit
    /* 116 */ SyscallDesc("gettimeofday", ignoreFunc), //32 bit

and gettimeofday() is declared in src/sim/syscal_emul.hh

/// Target gettimeofday() handler.
template <class OS>
SyscallReturn
gettimeofdayFunc(SyscallDesc *desc, int callnum, LiveProcess *process,
        ThreadContext *tc)
{
    TypedBufferArg<typename OS::timeval> tp(tc->getSyscallArg(0));

    getElapsedTime(tp->tv_sec, tp->tv_usec);
    tp->tv_sec += seconds_since_epoch;
    tp->tv_sec = htog(tp->tv_sec);
    tp->tv_usec = htog(tp->tv_usec);

    tp.copyOut(tc->getMemPort());

    return 0;
}

Does anybody know how to solve this problem.

for other benchmarks, error are getting because of unimplemented
instructions like for, fpadd32s
i found out these should be implemented in /arch/sparc/isa/decoder.isa
so, i checked in /arch/alpha/isa/decoder.isa to get an idea so that i can
implement in sparc, but for and fpadd32s are not implemented overthere.

Could anybody help me in this regard by give a brief idea on this.

Thanks,
Pavan


On Fri, Jan 30, 2009 at 12:26 PM, Ali Saidi <[email protected]> wrote:

>
> On Jan 30, 2009, at 1:11 PM, Pavan Kumar wrote:
>
> > Hello,
> >
> >          I am trying to run SPEC CPU 2006 benchmarks on
> > m5-2.0b6(SPARC_SE mode), and some of them are showing the following
> > error messages.
> >
> >
> > perlbench  ---           error (fatal: syscall gettimeofday (#116)
> > unimplemented.)
> You need to implement gettimeofday(). Implementations exist for
> AlphaLinux and Tru64. Replacing the unimplementedFunc in the syscall
> table with the appropriate one should do the trick.
>
> > gcc           ---           error (fatal: syscall stat (#38)
> > unimplemented.)
> Same as above
>
> > bwaves     ---           error (fatal: syscall ioctl (#54)
> > unimplemented.)
> Maybe same as above. You would need to figure out what it's calling
> ioctl on.
>
> >
> > gobmk      ---           error (fatal: fault (fp_disabled) detected
> > @ PC 0x16e74c)
> I'm aware of no programs that intentionally take an fp fault for
> correct behavior, but I could be wrong.
>
> >
> > soplex      ---           error (panic: attempt to execute
> > unimplemented instruction 'for' (inst 0x7c99b30f88))
> Implement the for instruction in src/arch/sparc/isa
>
> >
> > povray      ---           error (fatal: syscall getcwd (#119)
> > unimplemented.)
> see above
>
> >
> > hmmer     ---           error (fatal: syscall gettimeofday (#116)
> > unimplemented.)
> see above
>
> > lbm          ---           error (fatal: syscall stat (#38)
> > unimplemented.)
> see above
>
> > astar        ---           error (panic: attempt to execute
> > unimplemented instruction 'fpadd32s' (inst 0x5385b18a63))
> Implement the fpadd32s instruction in src/arch/sparc/isa
>
> > xalancbmk---           error (fatal: syscall gettimeofday (#116)
> > unimplemented.)
> see above
>
> >
> > zeusmp    ---           error (fatal: syscall ioctl (#54)
> > unimplemented.)
> see above
>
> If you're going to work on getting these fixed, please use the
> development repository http://repo.m5sim.org/m5 and PLEASE contribute
> the changes back when you are done.
>
> Thanks,
> Ali
>
> _______________________________________________
> m5-users mailing list
> [email protected]
> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
>
_______________________________________________
m5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/m5-users

Reply via email to