You're right, the final stage is not doing a compile, and takes only 1-2 seconds. I'm reverting this change.
On Mon, May 3, 2010 at 4:42 PM, Martin Bligh <[email protected]> wrote: > This looks very strange. Surely we *do* want to call kernel.clean? > If that is doing a compile, I'd say the callee needs fixing, not the > caller? > > > On Mon, May 3, 2010 at 3:59 PM, Duane Sand <[email protected]> wrote: > > Eliminate useless final kernel compile from cleanup phase of > > kernbench. Kernbench with iterations=N does N=2 compilations: > > 1 cold compile from disk and get sources cached in filesystem ram; > > then N timed warm compilations for the measurements; then 1 mistaken > > final compilation that was intended just to delete output files > > but instead triggered a complete compilation. > > > > Signed-off-by: Duane Sand <[email protected]> > > > > --- autotest/client/tests/kernbench/kernbench.py 2010-05-03 > 15:30:40.000000000 -0700 > > +++ autotest/client/tests/kernbench/kernbench.py 2010-05-03 > 15:30:40.000000000 -0700 > > @@ -66,10 +66,6 @@ > > self.kernel.build_timed(self.threads, self.timefile) > > > > > > - def cleanup(self): > > - self.kernel.clean(logged=False) # Don't leave litter lying > around > > - > > - > > def postprocess_iteration(self): > > os.chdir(self.resultsdir) > > utils.system("grep -h elapsed %s >> time" % self.timefile) > > _______________________________________________ > > Autotest mailing list > > [email protected] > > http://test.kernel.org/cgi-bin/mailman/listinfo/autotest > > >
_______________________________________________ Autotest mailing list [email protected] http://test.kernel.org/cgi-bin/mailman/listinfo/autotest
