[PATCH] perf: fix use_browser infinite loop bug in annotate

[PATCH] perf: fix use_browser infinite loop bug in annotate

[Stephane Eranian]

This patch fixes an infinite loop problem with perf annotate when
the TUI is not used. When the perfconfig file is not found use_browser
remains at -1, when found and tui is disabled use_browser = 0. Thus the
correct test for TUI is use_browser > 0.

Signed-off-by: Stephane Eranian <eranian@google.com>

diff --git a/tools/perf/builtin-annotate.c b/tools/perf/builtin-annotate.c
index 08278ed..96db524 100644
--- a/tools/perf/builtin-annotate.c
+++ b/tools/perf/builtin-annotate.c
@@ -343,7 +343,7 @@ find_next:
 			continue;
 		}
 
-		if (use_browser) {
+		if (use_browser > 0) {
 			key = hist_entry__tui_annotate(he);
 			if (is_exit_key(key))
 				break;

Re: [PATCH] perf: fix use_browser infinite loop bug in annotate

[Arnaldo Carvalho de Melo]

Em Thu, May 27, 2010 at 02:50:01PM +0200, Stephane Eranian escreveu:
> This patch fixes an infinite loop problem with perf annotate when
> the TUI is not used. When the perfconfig file is not found use_browser
> remains at -1, when found and tui is disabled use_browser = 0. Thus the
> correct test for TUI is use_browser > 0.
> 
> Signed-off-by: Stephane Eranian <eranian@google.com>

Thanks, I'm applying this one and doing an audit of other uses, I know
of one more case reported by Frédéric that was fixed in a similar way.

> diff --git a/tools/perf/builtin-annotate.c b/tools/perf/builtin-annotate.c
> index 08278ed..96db524 100644
> --- a/tools/perf/builtin-annotate.c
> +++ b/tools/perf/builtin-annotate.c
> @@ -343,7 +343,7 @@ find_next:
>  			continue;
>  		}
>  
> -		if (use_browser) {
> +		if (use_browser > 0) {
>  			key = hist_entry__tui_annotate(he);
>  			if (is_exit_key(key))
>  				break;

Re: [PATCH] perf: fix use_browser infinite loop bug in annotate

[Arnaldo Carvalho de Melo]

Em Thu, May 27, 2010 at 11:51:42AM -0300, Arnaldo Carvalho de Melo escreveu:
> Em Thu, May 27, 2010 at 02:50:01PM +0200, Stephane Eranian escreveu:
> > This patch fixes an infinite loop problem with perf annotate when
> > the TUI is not used. When the perfconfig file is not found use_browser
> > remains at -1, when found and tui is disabled use_browser = 0. Thus the
> > correct test for TUI is use_browser > 0.
> > 
> > Signed-off-by: Stephane Eranian <eranian@google.com>
> 
> Thanks, I'm applying this one and doing an audit of other uses, I know
> of one more case reported by Frédéric that was fixed in a similar way.

	Oops, I already have this fixed on changeset

62e3436b5f3461662929eae102beefbd12127cb1

	That is at:

git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux-2.6 perf

http://git.kernel.org/?p=linux/kernel/git/acme/linux-2.6.git;a=commit;h=62e3436b5f3461662929eae102beefbd12127cb1
 
	I pushed this to Ingo yesterday, should be in tip/perf/core soon,
thanks anyway!

- Arnaldo

Re: [PATCH] perf: fix use_browser infinite loop bug in annotate

[Stephane Eranian]

On Thu, May 27, 2010 at 5:45 PM, Arnaldo Carvalho de Melo
<acme@infradead.org> wrote:
> Em Thu, May 27, 2010 at 11:51:42AM -0300, Arnaldo Carvalho de Melo escreveu:
>> Em Thu, May 27, 2010 at 02:50:01PM +0200, Stephane Eranian escreveu:
>> > This patch fixes an infinite loop problem with perf annotate when
>> > the TUI is not used. When the perfconfig file is not found use_browser
>> > remains at -1, when found and tui is disabled use_browser = 0. Thus the
>> > correct test for TUI is use_browser > 0.
>> >
>> > Signed-off-by: Stephane Eranian <eranian@google.com>
>>
>> Thanks, I'm applying this one and doing an audit of other uses, I know
>> of one more case reported by Frédéric that was fixed in a similar way.
>
>        Oops, I already have this fixed on changeset
>
> 62e3436b5f3461662929eae102beefbd12127cb1
>
>        That is at:
>
> git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux-2.6 perf
>
> http://git.kernel.org/?p=linux/kernel/git/acme/linux-2.6.git;a=commit;h=62e3436b5f3461662929eae102beefbd12127cb1
>
>        I pushed this to Ingo yesterday, should be in tip/perf/core soon,
> thanks anyway!
>
Ok, if it's fixed then this is fine. The other problem I reported is
more important.
Thanks for checking.