Hello, I intend to observe all the physical DRAM addresses that are accessed when an application is being executed in full-system mode. Following some earlier discussions on this mailing list, I used the commMonitor. I added the following lines to fs.py file: test_sys.monitor = CommMonitor() test_sys.monitor.trace = MemTraceProbe(trace_file = "monitor.ptrc.gz")
Next, I connected the CommMonitor as follows in the CacheConfig.py file: system.l2.mem_side = system.monitor.slave system.monitor.master = system.membus.slave I create and then restore from a checkpoint just before the application is started. Following is my command line: ./build/X86/gem5.opt configs/example/fs.py --kernel /home/myComputer/Desktop/x86_64-vmlinux-2.6.28.4-smp --disk-image /home/myComputer/Desktop/x86root-parsec.img --script /home/myComputer/Desktop/gem5/configs/boot/test.rcS --caches --l2cache --checkpoint-restore=1 --restore-with-cpu=TimingSimpleCPU Next, I use the decode_packet_trace.py file in utils folder to process the .ptrc file obtained after simulation. Following is a sample of the output of this step: 5,u,531425792,64,256,4986860244500 6,u,535891200,64,2,4986860333000 9,u,529653760,64,512,4986868293000 9,u,530137088,64,512,4986868293000 9,u,529657856,64,512,4986868293000 9,u,530022400,64,512,4986868293000 10,u,531425216,64,256,4986868293000 8,u,529655744,64,512,4986868294500 8,u,521711552,64,512,4986868294500 ... The 3rd column in this output seems to be the address. However, DRAM addresses are usually represented in Hexadecimal format. In the entire output file, I did not find any alphabet character. All addresses are just numbers. Is this expected output? Do I need to convert these addresses to hexadecimal format manually next or did I make a mistake in the full-sytem setup? Thanks, Biresh
_______________________________________________ gem5-users mailing list -- gem5-users@gem5.org To unsubscribe send an email to gem5-users-le...@gem5.org %(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s
