You're the first person I know of to try and run spec2006 on M5.
(1) if you ran a Tru64 binary ftruncate would probably work just fine,
but the corresponding entry isn't in the linux syscall table. Adding
it might make things work, but the parameters may be different.
Changing ftruncate in src/arch/alpha/linux/process.cc from
unimplementedFunc to ftruncate will probably fix the problem.
(2) mremap isn't implemented, but you can certainly implement it. Just
use the implementation of mmap as a guide.
(3) You can probably implement this by either just copying some dummy
value into the buffer, or simply passing the call to the real os like
we do in most places.
(4) I don't know if perlbench has an unaligned access intentionally or
not (e.g. is there something prior to it that we did wrong that caused
the fault, or does it actually happen). If it actually happens you
would need to handle unaligned faults in SE mode. This wouldn't be too
hard. You could look at how we take TLB faults in SE for a starting
point, but pretty much when the fault happens just do two reads to
grab the data in an aligned fashion, parse the faulting instruction to
find the destination register, write it there and skip the faulting
instruction (that is what happens in the kernel).
(5) Did you increase the size of the memory allocated by the Physical
Memory object?
If you fix any of these please send us patches.
Ali
On Apr 15, 2008, at 9:39 PM, Meng-Ju Wu wrote:
Hi all,
I am working on SPEC2006 benchmarks and I got five kinds of errors on
M5 2.0b5 SE mode.
(1) fatal: syscall ftruncate (#130) unimplemented. (Example:
410.bwaves )
(2) fatal: syscall mremap (#341) unimplemented. (Example: 429.mcf )
(3) fatal: syscall getcwd (#367) unimplemented. (Example: 453.povray)
(4) fatal: fault (unalign) detected @ PC 0x12009cedc (Example:
400.perlbench)
(5) terminate called after throwing an instance of 'std::bad_alloc'
(Example: 470.lbm)
For (1), I remembered that ftruncate() was implemented in M5 1.0. I
check the M5 2.0b5 code, and it is only implemented in tru64.
For (5), I tried to increase the memory size, but it is not helpful.
I am really appreciate if you can give me some hints to solve
(1)~(5).
Thanks,
Meng-Ju
_______________________________________________
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
