Hi, I've got two comments. First, not all ISAs put their system call implementations in a process.cc file. Some (including x86) have them split out into a syscalls.cc.
Second, this is actually an interesting coincidence because I was just about to start looking at getting spec2006 to work better. A guy named Vince Weaver did some work testing our x86 implementation a while ago, and he has some patches that haven't yet made it into the code base. You can find those here: http://www.csl.cornell.edu/~vince/projects/m5/ I've heard that some of these patches help getting spec2006 to run, specifically with the system call issues but perhaps in other ways as well. One of the things I'm planning to do while working on this is to integrate those patches into the code base. Gabe On 02/28/11 09:54, Korey Sewell wrote: > What CPU model are you running these on? > > My first reaction is for the unimplemented syscalls, you can take a > look at how the other ISAs implemented the same syscall and add your > new version (and if you do, submit the patch to M5 reviewboard!). > > Check, src/arch/<isa>/<os>/process.cc for the list of system calls. > The definitions should be in that file or maybe src/sim/sycall_emul.hh/cc. > > For the unmapped address, that's problem an bug in ISA specification > that needs to be fixed. Usually, a bad value is propagated through the > registers somewhere when that happens (in my experience). If you run > the execution trace on (--trace-flags=Exec), you may be able to find > the faulty instruction and that can help someone help debug it for you. > > On Mon, Feb 28, 2011 at 12:43 PM, Andrea Pellegrini > <[email protected] <mailto:[email protected]>> wrote: > > Andrea Pellegrini <apellegr <at> umich.edu <http://umich.edu>> writes: > > > > > > > Hi all, > > I saw an old post on the M5 mailing list and I was wondering if > anybody > > was able to run SPEC2006 on the standalone x86 version of M5. I > compiled > > them with the "-static" option, but I am still having issues > running them > > in the > > simulator. > > If anybody was successful, which are the settings that allowed > you to do > > so? > > > > Thanks, > > -Andrea > > > > > Turns out that most benchmarks run out of the box. For 401.bzip2 I > had to add > extra physical memory. > > I still have issues with some of them. In brief, how can I fix: > > 400.perlbench - fatal: syscall ioctl (#16) unimplemented. > 473.astar - panic: Tried to read unmapped address 0. > 483.xalancbmk - fatal: syscall lstat (#6) unimplemented. > > Thanks! > -Andrea > > More in detail: > > > =========================================================================== > 400.perlbench: > **** REAL SIMULATION **** > info: Entering event queue @ 0. Starting simulation... > warn: instruction 'fnstcw_Mw' unimplemented > For more information see: http://www.m5sim.org/warn/437d5238 > warn: instruction 'fldcw_Mw' unimplemented > For more information see: http://www.m5sim.org/warn/437d5238 > fatal: syscall ioctl (#16) unimplemented. > @ cycle 78692500 > [unimplementedFunc:build/X86_SE/sim/syscall_emul.cc, line 82] > Memory Usage: 2190824 KBytes > For more information see: http://www.m5sim.org/fatal/be2b42f > > =========================================================================== > 471.omnetpp: > **** REAL SIMULATION **** > info: Entering event queue @ 0. Starting simulation... > warn: instruction 'fnstcw_Mw' unimplemented > For more information see: http://www.m5sim.org/warn/437d5238 > warn: instruction 'fldcw_Mw' unimplemented > For more information see: http://www.m5sim.org/warn/437d5238 > OMNeT++ Discrete Event Simulation (C) 1992-2003 Andras Varga > See the license for distribution terms and warranty disclaimer > info: Increasing stack size by one page. > <!> Cannot open `omnetpp.ini'. > > It's a problem with the benchmark itself, I should be able to fix > it quickly. > > =========================================================================== > 473.astar: > > **** REAL SIMULATION **** > info: Entering event queue @ 0. Starting simulation... > warn: instruction 'fnstcw_Mw' unimplemented > For more information see: http://www.m5sim.org/warn/437d5238 > warn: instruction 'fldcw_Mw' unimplemented > For more information see: http://www.m5sim.org/warn/437d5238 > info: Increasing stack size by one page. > Small Path Finding Library > Read configuration file > Create ways > BigLakes2048.bin > panic: Tried to read unmapped address 0. > @ cycle 22677000 > [invoke:build/X86_SE/arch/x86/faults.cc, line 290] > Memory Usage: 2188136 KBytes > For more information see: http://www.m5sim.org/panic/d71dd0bc > Program aborted at cycle 22677000 > > =========================================================================== > 483.xalancbmk: > > **** REAL SIMULATION **** > info: Entering event queue @ 0. Starting simulation... > warn: instruction 'fnstcw_Mw' unimplemented > For more information see: http://www.m5sim.org/warn/437d5238 > warn: instruction 'fldcw_Mw' unimplemented > For more information see: http://www.m5sim.org/warn/437d5238 > info: Increasing stack size by one page. > info: Increasing stack size by one page. > info: Increasing stack size by one page. > fatal: syscall lstat (#6) unimplemented. > @ cycle 819709500 > [unimplementedFunc:build/X86_SE/sim/syscall_emul.cc, line 82] > Memory Usage: 2210696 KBytes > For more information see: http://www.m5sim.org/fatal/be2b42f > > > _______________________________________________ > m5-users mailing list > [email protected] <mailto:[email protected]> > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users > > > > > -- > - Korey > > > _______________________________________________ > 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
