Re: [PATCH 1/6] perf, tools, stat: Abstract stat metrics printing

[Jiri Olsa <jolsa@redhat.com>]

On Tue, Dec 22, 2015 at 02:16:30AM +0100, Andi Kleen wrote:
> > > -			fprintf(out, "                                   ");
> > > +			print_metric(ctxp, NULL, NULL, "insn per cycle", 0);
> > >  		}
> > >  		total = avg_stats(&runtime_stalled_cycles_front_stats[ctx][cpu]);
> > >  		total = max(total, avg_stats(&runtime_stalled_cycles_back_stats[ctx][cpu]));
> > >  
> > > +		out->new_line(ctxp);
> > >  		if (total && avg) {
> > >  			ratio = total / avg;
> > > -			fprintf(out, "\n");
> > 
> > you haven't address my first comment in here http://marc.info/?l=linux-kernel&m=144662610723134&w=2
> 
> The new_line is always needed because stalled cycles is always printed.
> 
> The reason it is always printed is that metric-only needs to see all the
> metrics for its column headers. That's why there are else cases
> everywhere.

please notice extra line below - between instructions and cycles lines

[jolsa@krava perf]$ ./perf stat -e instructions,cycles  kill
kill: not enough arguments

 Performance counter stats for 'kill':

           769,784      instructions              #    0.78  insn per cycle         
                                                                                    
           990,795      cycles                                                      


jirka