A correction in the command to take the checkpoint, the command used to
take a checkpoint is:

./build/ALPHA_MOESI_hammer/gem5.fast configs/spec2006/spec2006.py astar
astar astar astar astar astar astar astar -n 8 --cpu-type=timing
--l1i_size=32kB --l1d_size=32kB --l2_size=2MB --num-l2caches=32 --caches
--l2cache --topology=MeshDirCorners --mesh-rows=2 --num-dirs=4
--checkpoint-at-end --checkpoint-dir=/export/home/tmp/ckpts/SE -I 10000



On Thu, Nov 1, 2012 at 5:30 PM, Jagadish Kotra <[email protected]>wrote:

> Hello,
>
>    ALPHA InorderCPU works fine (without checkpointing) in SE mode with
> Ruby. However, when I try to restore the checkpoint taken with TimingCPU
> (using MOESI_hammer ruby protocol) and restore it with Timing CPU but
> switch to InorderCPU I get a SEGFault. Here are the commands I am using to
> create/restore the checkpoints...
>
> Taking a checkpoint:
>
> ./build/ALPHA_MOESI_CMP_directory/gem5.debug configs/spec2006/spec2006.py
> astar -n 8 --restore-with-cpu=timing --cpu-type=inorder --l1i_size=32kB
> --l1d_size=32kB --l2_size=2MB --num-l2caches=8 --caches --l2cache
> --topology=MeshDirCorners --mesh-rows=2 --num-dirs=4 -r 2
> --checkpoint-dir=/export/home/tmp/ckpts/SE -I 100000
>
> Restoring a checkpoint:
>
>  gdb --args ./build/ALPHA_MOESI_CMP_directory/gem5.debug
> --debug-flags=InOrderCPU,Resource  configs/spec2006/spec2006.py astar astar
> astar astar astar astar astar astar -n 8 --restore-with-cpu=timing
> --cpu-type=inorder --l1i_size=32kB --l1d_size=32kB --l2_size=2MB
> --num-l2caches=8 --caches --l2cache --topology=MeshDirCorners --mesh-rows=2
> --num-dirs=4 -r 2 --checkpoint-dir=/export/home/tmp/ckpts/SE -I 100000
>
> Here is the output:
>
> Benchmarks/benchspec/CPU2006/473.astar/exe/astar_base.amd64-m64-gcc42-nn
> warn: CoherentBus system.membus has no snooping ports attached!
>     *  *0: system.switch_cpus0.ResourcePool: Initializing resource:
> system.switch_cpus0.fetch_seq_unit.
>       0: system.switch_cpus0.ResourcePool: Initializing resource:
> system.switch_cpus0.icache_port.*
> *
> Program received signal SIGSEGV, Segmentation fault.
> 0x00000000005bd7a2 in MasterPort::peerBlockSize (this=0x2b417c8)
>     at build/ALPHA_MOESI_CMP_directory/mem/port.cc:161
> 161         return _slavePort->deviceBlockSize();
> Missing separate debuginfos, use: debuginfo-install
> glibc-2.12-1.80.el6_3.5.x86_64 keyutils-libs-1.4-4.el6.x86_64
> krb5-libs-1.9-33.el6_3.2.x86_64 libcom_err-1.41.12-12.el6.x86_64
> libgcc-4.4.6-4.el6.x86_64 libselinux-2.0.94-5.3.el6.x86_64
> libstdc++-4.4.6-4.el6.x86_64 openssl-1.0.0-25.el6_3.1.x86_64
> python-2.6.6-29.el6_3.3.x86_64 python-libs-2.6.6-29.el6_3.3.x86_64
> zlib-1.2.3-27.el6.x86_64
> (gdb) bt
> #0  0x00000000005bd7a2 in MasterPort::peerBlockSize (this=0x2b417c8)
>     at build/ALPHA_MOESI_CMP_directory/mem/port.cc:161
> #1  0x0000000000ad9fb8 in CacheUnit::init (this=0x2b41fd0)
>     at
> build/ALPHA_MOESI_CMP_directory/cpu/inorder/resources/cache_unit.cc:115
> #2  0x0000000000b30620 in ResourcePool::init (this=0x2b41de0)
>     at build/ALPHA_MOESI_CMP_directory/cpu/inorder/resource_pool.cc:140
> #3  0x0000000000b56247 in InOrderCPU::init (this=0x2b41070)
>     at build/ALPHA_MOESI_CMP_directory/cpu/inorder/cpu.cc:808
> #4  0x0000000001165d8f in _wrap_SimObject_init (args=0x2915910)
>     at
> build/ALPHA_MOESI_CMP_directory/python/m5/internal/param_SimObject_wrap.cc:3005
> #5  0x0000003c3d0deb24 in PyEval_EvalFrameEx ()
>    from /usr/lib64/libpython2.6.so.1.0
> #6  0x0000003c3d0e0797 in PyEval_EvalCodeEx ()
>    from /usr/lib64/libpython2.6.so.1.0
> #7  0x0000003c3d0debe4 in PyEval_EvalFrameEx ()
>    from /usr/lib64/libpython2.6.so.1.0
> #8  0x0000003c3d0e0797 in PyEval_EvalCodeEx ()
>    from /usr/lib64/libpython2.6.so.1.0
>
> I have traced the code and it seems for some reason the _slavePort (in
> mem/port.cc) corresponding to the iCachePort is becoming NULL and hence the
> call to know the cacheBlockSize results in a SEG Faults. I have further
> used --debug-flags=InOrder,Resource to see the problem and I could see the
> iCachePort initialized successfully!
> Not sure if InorderCPU is supports checkpoint restoration as yet!
>
> I am using the following gem5 revision: 9321:564042b95388.
>
> Can someone please let me know on how can this be fixed?
>
> Appreciate your help!
>
> Regards,
> Jagadish.
>
_______________________________________________
gem5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users

Reply via email to