----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: http://reviews.gem5.org/r/1021/#review2051 -----------------------------------------------------------
Overall this looks worthwhile, thanks for submitting it. I hope you have the time to fix up the minor issues I spotted. util/tracediff <http://reviews.gem5.org/r/1021/#comment2537> Shouldn't this default to "." not ""? Otherwise it looks like the code below will place the trace files in the root dir rather than the current dir if -p isn't specified. util/tracediff <http://reviews.gem5.org/r/1021/#comment2538> I think this code breaks if you specify both -p and -n in that order (it'll pick up the -p and not the -n). That may not be a common combination but it would be nice to get it right anyway. Seems like we really need a while loop that checks for $ARGV[0][0] eq '-' or something like that (or use the perl equivalent of getopt or parseargs or something, but that could be overkill). It's been a long time since I've done much perl so I can't be more specific than that. - Steve Reinhardt On Jan. 30, 2012, 3:22 a.m., William Wang wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > http://reviews.gem5.org/r/1021/ > ----------------------------------------------------------- > > (Updated Jan. 30, 2012, 3:22 a.m.) > > > Review request for Default. > > > Description > ------- > > DEBUG: Added a path option to tracediff > > Added a path option '-p' to tracediff, this is useful when the default cwd > quota/space is not large enough to hold all the traces. > > > Diffs > ----- > > util/tracediff 0a742249f76b > > Diff: http://reviews.gem5.org/r/1021/diff/diff > > > Testing > ------- > > Ran tracediff okay: ./util/tracediff -p ~/scratch 'build/ARM_FS| > build/ARM_FS_PASS#/gem5.opt' --debug-flags=Exec,Cache,Bus > --trace-start=66488126500 -re tests/run.py > build/ARM_FS/tests/opt/long/10.linux-boot/arm/linux/realview-o3-dual > > > Thanks, > > William Wang > > _______________________________________________ gem5-dev mailing list [email protected] http://m5sim.org/mailman/listinfo/gem5-dev
