Basically, the problem was that a form of Nop wasn't being decoded 
correctly which threw off the location of instruction boundaries. It's 
fixed in the repository.

Gabe

On Tue, 29 Jul 2008, Viren Kumar wrote:

> Hello everyone,
>
> I've just started using M5 in X86 SE mode. My ultimate goal is to run some
> SPEC 2000 benchmarks on it, so I thought I'd start by compiling the provided
> Hello World application myself to test the waters.
>
> First, I built m5.debug and that was fine. I then compiled the Hello World
> program with gcc 4.1.2 on an AMD64 Barcelona system running Gentoo 2.6.21
>
> barcelona# gcc -o test test.c --static
>
> I compared this against the given /tests/test-progs/hello/bin/x86/linux/hello
> and saw the difference in kernel versions with file:
>
> $ file ../tests/test-progs/hello/bin/x86/linux/hello
> ../tests/test-progs/hello/bin/x86/linux/hello: ELF 64-bit LSB executable,
> x86-64, version 1 (SYSV), for GNU/Linux 2.4.1, statically linked, not
> stripped
>
> and mine was:
>
> $ file test
> test: ELF 64-bit LSB executable, x86-64, version 1 (SYSV), for GNU/Linux
> 2.6.9, statically linked, not stripped
>
> I changed the command option in configs/example/se.py and ran the script with
> tracing in m5 and got:
>
> $ build/X86_SE/m5.debug --trace-flags="X86" configs/example/se.py
> ...
> ...
>  36500: global: Setting int reg 17 to value 0x60
>  36500: global: Setting int reg 32 to value 0
>  37000: global: Setting int reg 16 to value 0
>  37000: global: Read int reg 16 and got value 0
>  37000: global: Read int reg 0 and got value 0x7fffffffee38
>  37000: global: Read int reg 17 and got value 0x60
>  37000: global: ADD_M_R : st: The address is 0x7fffffffedc3
>  37500: global: Setting int reg 16 to value 0
> fatal: fault (Invalid-Opcode) detected @ PC 0x400332
> @ cycle 37500
> [invoke:build/X86_SE/sim/faults.cc, line 43]
> Memory Usage: 543424 KBytes
>
>
> I'm not sure what's causing the Invalid Opcode problem here. I searched the
> mailing list archive but wasn't able to find anything on this error, so
> thought I'd post here and ask.
>
> Thanks in advance,
> Viren
>
>
>
> _______________________________________________
> 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