Just run m5.debug under gdb, and when you hit the segfault, type 'where' at the gdb prompt.
On Sun, Mar 21, 2010 at 9:33 PM, Ashutosh Jain <[email protected]> wrote: > Hi > How can I do stack backtrace in M5? > > On Sun, Mar 21, 2010 at 11:32 PM, Steve Reinhardt <[email protected]> wrote: >> >> OK; can you provide a stack backtrace for the segfault? >> >> On Sun, Mar 21, 2010 at 9:25 PM, Ashutosh Jain <[email protected]> >> wrote: >> > Hi >> > Yes I have compiled the benchmarks and got the binaries.. >> > Ashutosh >> > >> > On Sun, Mar 21, 2010 at 9:45 PM, Steve Reinhardt <[email protected]> >> > wrote: >> >> >> >> We'd need a stack backtrace from the segfault to begin to guess what's >> >> going on... >> >> >> >> Where did you get the binaries from? Did you compile them yourself? >> >> >> >> Steve >> >> >> >> On Sun, Mar 21, 2010 at 4:53 PM, Ashutosh Jain <[email protected]> >> >> wrote: >> >> > Hi >> >> > I am working on the updated M5 from the repository. I would like to >> >> > run >> >> > simulations with SPEC CPU 2006 benchmarks; but every >> >> > time the simulation >> >> > ends with segmentation fault. Is this the problem with M5 or SPEC CPU >> >> > 2006 >> >> > or am I missing something. >> >> > >> >> > Consider the case of benchmark bzip2. I have set it like: >> >> > ------------------- >> >> > specbzip2 = os.path.join(test_base, >> >> > '/home/ashutoshj/newm5/configs/cpu2006/speccpu2006/bzip2/') >> >> > specbzip2out = os.path.join(test_base, >> >> > '/home/ashutoshj/newm5/configs/cpu2006/speccpu2006/bzip2/') >> >> > class SPECBZIP2(LiveProcess): >> >> > executable = os.path.join(specbzip2, 'bzip2_peak.alpha') >> >> > cmd = 'bzip2_peak.alpha input.source 280 >> >> > output = os.path.join(specbzip2out, 'input.source.out') >> >> > --------------------------------- >> >> > And used in the se.py file. Then I ran the simulation it gives the >> >> > output >> >> > as: >> >> > ---------------------------------------------- >> >> > M5 Simulator System >> >> > Copyright (c) 2001-2008 >> >> > The Regents of The University of Michigan >> >> > All Rights Reserved >> >> > >> >> > M5 compiled Mar 10 2010 06:42:32 >> >> > M5 revision 79463f723460+ 6018+ default tip >> >> > M5 started Mar 21 2010 19:31:14 >> >> > M5 executing on leif.cs.lamar.edu >> >> > command line: ../../build/ALPHA_SE/m5.opt -d cmp01_thds01_L2cache01 >> >> > cmp01_thds01_L2cache01.py --detailed --caches >> >> > max_insts_all_threads = 100000 >> >> > SPECBZIP2() >> >> > system.cpu[0].workload = Benchmarks.SPECBZIP2() >> >> > Global frequency set at 1000000000000 ticks per second >> >> > Segmentation fault >> >> > --------------------------------------------------- >> >> > Also all the SPEC CPU 2006 benchmarks with their inputs, outputs, and >> >> > executables in their respective directories are present at : >> >> > /home/ashutoshj/newm5/configs/cpu2006/speccpu2006/ >> >> > >> >> > With exactly the same configurations, SPEC CPU2000 gave me the >> >> > desired >> >> > expected results; So I am unable to find a solution to this problem. >> >> > Hoping to hear a solution >> >> > -- >> >> > Ashutosh Jain >> >> > >> >> > >> >> > >> >> > _______________________________________________ >> >> > 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 >> > >> > >> > >> > -- >> > Ashutosh Jain >> > >> > >> > >> > _______________________________________________ >> > 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 > > > > -- > Ashutosh Jain > > > > _______________________________________________ > 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
