Re: [PATCH] perf hists browser: Fix segfault if use symbol filter in cmdline

[Arnaldo Carvalho de Melo]

Em Thu, Dec 03, 2015 at 12:49:42AM +0900, Namhyung Kim escreveu:
> Hi,
> 
> On Wed, Dec 02, 2015 at 11:28:52AM +0000, Wang Nan wrote:
> > If feed perf a symbol filter in cmdline and the result is empty,
> > pressing 'Enter' in the hist browser causes crash:
> > 
> >  # ./perf report perf.data   <-- Common mistake for beginners
> > 
> > Then press 'Enter':
> > 
> >  perf: Segmentation fault
> >  -------- backtrace --------
> >  /home/wangnan/perf[0x53e578]
> >  /lib64/libc.so.6(+0x3545f)[0x7f76bafe045f]
> >  /home/wangnan/perf[0x539dd4]
> >  /home/wangnan/perf(perf_evlist__tui_browse_hists+0x96)[0x53d216]
> >  /home/wangnan/perf(cmd_report+0x1b9f)[0x442c7f]
> >  /home/wangnan/perf[0x47efa2]
> >  /home/wangnan/perf(main+0x5f5)[0x432fa5]
> >  /lib64/libc.so.6(__libc_start_main+0xf4)[0x7f76bafccbd4]
> >  /home/wangnan/perf[0x4330d4]
> > 
> > This is because 'perf.data' is interperted as a symbol filter, and the
> > result is empty, so selection is empty. However,
> > hist_browser__toggle_fold() forgets to check it.
> > 
> > This patch simply return faluse when selection is NULL.
> 
> s/faluse/false/

Fixed this and the 'interperted' and some other text details, and also
the container_of usage as reported by you below.

Thanks!

- Arnaldo
 
> > 
> > Signed-off-by: Wang Nan <wangnan0@xxxxxxxxxx>
> > Cc: Arnaldo Carvalho de Melo <acme@xxxxxxxxxx>
> 
> Acked-by: Namhyung Kim <namhyung@xxxxxxxxxx>
> 
> One nitpick below..
> 
> > ---
> >  tools/perf/ui/browsers/hists.c | 5 ++++-
> >  1 file changed, 4 insertions(+), 1 deletion(-)
> > 
> > diff --git a/tools/perf/ui/browsers/hists.c b/tools/perf/ui/browsers/hists.c
> > index dcdcbaf..33da341 100644
> > --- a/tools/perf/ui/browsers/hists.c
> > +++ b/tools/perf/ui/browsers/hists.c
> > @@ -337,9 +337,12 @@ static bool hist_browser__toggle_fold(struct hist_browser *browser)
> >  {
> >  	struct hist_entry *he = browser->he_selection;
> >  	struct map_symbol *ms = browser->selection;
> > -	struct callchain_list *cl = container_of(ms, struct callchain_list, ms);
> > +	struct callchain_list *cl = ms ? container_of(ms, struct callchain_list, ms) : NULL;
> 
> I think this line is not needed as container_of doesn't make a pointer
> dereference by itself.  Just checking below is enough IMHO.

Agreed, changing this.
 
> Thanks,
> Namhyung
> 
> 
> >  	bool has_children;
> >  
> > +	if (!ms)
> > +		return false;
> > +
> >  	if (ms == &he->ms)
> >  		has_children = hist_entry__toggle_fold(he);
> >  	else
> > -- 
> > 1.8.3.4
> > 
--
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/