Re: [PATCH] perf/powerpc: Cache the DWARF debug info

[Jiri Olsa]

On Thu, Oct 23, 2014 at 12:33:22PM -0300, Arnaldo Carvalho de Melo wrote:
> Em Thu, Oct 23, 2014 at 05:13:06PM +0200, Jiri Olsa escreveu:
> > On Thu, Oct 23, 2014 at 11:26:34AM -0300, Arnaldo Carvalho de Melo wrote:
> > > That is why I thought it would be a compromise to put what he did, it
> > > would not make the existing situation that much worse, work needs to be
> > > done in this area :-\
>  
> > I think we just need to put that libdw handle into dso object
> > as I suggested above
> 
> Isn't it there already?
> 
> The patch does:
> 
> +++ b/tools/perf/util/dso.h
> @@ -127,6 +127,7 @@ struct dso {
>         const char       *long_name;
>         u16              long_name_len;
>         u16              short_name_len;
> +       void            *dwfl;                  /* DWARF debug info */
> 
> 
> ----------
> 
> What you want, on top of that, at a minimum, we somehow limit the number
> of simultaneously dwfl_begin()'ed DSOs, right?

no, the patch is doing that.. as I wrote:

---
I think this can stay in 'struct dso' which is limited
by that code you show below (dso__data_open code)

I think we need just add generic functions that allocates/destroys
the dwfl handle and lazy allocate&store this handle whenever it's
needed and destroy it in dso__delete
---

so just:
  1) generic function to do the lazy allocation of dwlf handle
  2) store the handle within dso object
  3) destroy it in dso__delete

the limitation for opened dso objects number should cover
the dwlf limit

thanks,
jirka
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/