Re: [PATCH 5/5] perf diff: Fix -o/--order option behavior

[Namhyung Kim <namhyung@kernel.org>]

On Sun, Jan 04, 2015 at 07:26:56PM +0100, Jiri Olsa wrote:
> On Sat, Dec 27, 2014 at 02:06:33PM +0900, Namhyung Kim wrote:
> 
> SNIP
> 
> > +		return 0;
> > +
> > +	fmt = zalloc(sizeof(*fmt));
> > +	if (fmt == NULL) {
> > +		pr_err("Memory allocation failed\n");
> > +		return -1;
> > +	}
> > +
> > +	fmt->cmp      = hist_entry__cmp_nop;
> > +	fmt->collapse = hist_entry__cmp_nop;
> > +
> > +	switch (compute) {
> > +	case COMPUTE_DELTA:
> > +		fmt->sort = hist_entry__cmp_delta_idx;
> > +		break;
> > +	case COMPUTE_RATIO:
> > +		fmt->sort = hist_entry__cmp_ratio_idx;
> > +		break;
> > +	case COMPUTE_WEIGHTED_DIFF:
> > +		fmt->sort = hist_entry__cmp_wdiff_idx;
> > +		break;
> > +	default:
> > +		BUG_ON(1);
> > +	}
> > +
> > +	list_add(&fmt->sort_list, &perf_hpp__sort_list);
> > +	return 0;
> 
> so the first 'fmt' which gets to sorting is the one for
> data__files[sort_idx] file, that sounds good..
> 
> but as the sorting goes through all the perf_hpp__sort_list list,
> it will hit the 'sort_idx' data again.. should you disable sort
> function for its 'fmt' then?

Do you really think it's needed?

Yes, it'll be called twice but I think it's a relatively rare case as
most entries will be sorted by the sort_idx and baseline columns.

Thanks,
Namhyung