Re: buggy perf callgraph output

[Frederic Weisbecker <fweisbec@gmail.com>]

On Thu, Dec 23, 2010 at 02:08:20PM +0100, Ingo Molnar wrote:
> 
> * Frederic Weisbecker <fweisbec@gmail.com> wrote:
> 
> > On Fri, Dec 17, 2010 at 04:32:58PM +1100, Nick Piggin wrote:
> > > That would be great, thanks! It's a really awesome tool, I appreciate
> > > your effort on it.
> > 
> > Thanks :)
> > 
> > So the below patch should fix the issue. Tell me if you encounter it
> > again, or anything else.
> > 
> > diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c
> > index d503670..c749ba6 100644
> > --- a/tools/perf/util/hist.c
> > +++ b/tools/perf/util/hist.c
> > @@ -356,7 +356,7 @@ static size_t ipchain__fprintf_graph_line(FILE *fp, int depth, int depth_mask,
> >  
> >  static size_t ipchain__fprintf_graph(FILE *fp, struct callchain_list *chain,
> >  				     int depth, int depth_mask, int period,
> > -				     u64 total_samples, int hits,
> > +				     u64 total_samples, u64 hits,
> >  				     int left_margin)
> 
> Hmm ... nice fix but why didnt GCC warn about the loss of half of the 'cumul' bits 
> in this call:
> 
>                         ret += ipchain__fprintf_graph(fp, chain, depth,
>                                                       new_depth_mask, i++,
>                                                       new_total,
>                                                       cumul,
>                                                       left_margin);
> 
> We really, really want GCC to warn about this - it's such a common mistake. What 
> disabled that warning? Very weird.

No idea. I thought perf had that in its warning arsenal :)

> 
> Thanks,
> 
> 	Ingo