linux-kernel - Re: [PATCH] perf stat-display: Check if snprintf()'s fmt argument is NULL

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <5FADEC01265BADB6+20230803171126.134677-1-ye@kaige.org>
Date:   Fri,  4 Aug 2023 01:11:26 +0800
From:   Kaige Ye <ye@...ge.org>
To:     irogers@...gle.com
Cc:     acme@...nel.org, adrian.hunter@...el.com,
        alexander.shishkin@...ux.intel.com, jolsa@...nel.org,
        linux-kernel@...r.kernel.org, linux-perf-users@...r.kernel.org,
        mark.rutland@....com, mingo@...hat.com, namhyung@...nel.org,
        peterz@...radead.org, ye@...ge.org
Subject: Re: [PATCH] perf stat-display: Check if snprintf()'s fmt argument is NULL

On Wed, 2 Aug 2023 10:45:47 -0700 Ian Rogers <irogers@...gle.com> wrote:
> >
> > It is undefined behavior to pass NULL as snprintf()'s fmt argument.
> > Here is an example to trigger the problem:
> >
> >   $ perf stat --metric-only -x, -e instructions -- sleep 1
> >   insn per cycle,
> >   Segmentation fault (core dumped)
> >
> > With this patch:
> >
> >   $ perf stat --metric-only -x, -e instructions -- sleep 1
> >   insn per cycle,
> >   ,
> >
> > Signed-off-by: Kaige Ye <ye@...ge.org>
> > ---
> >  tools/perf/util/stat-display.c | 6 +++---
> >  1 file changed, 3 insertions(+), 3 deletions(-)
> >
> > diff --git a/tools/perf/util/stat-display.c b/tools/perf/util/stat-display.c
> > index 7329b3340..e8936cffd 100644
> > --- a/tools/perf/util/stat-display.c
> > +++ b/tools/perf/util/stat-display.c
> > @@ -438,7 +438,7 @@ static void print_metric_csv(struct perf_stat_config *config __maybe_unused,
> >                 fprintf(out, "%s%s", config->csv_sep, config->csv_sep);
> >                 return;
> >         }
> > -       snprintf(buf, sizeof(buf), fmt, val);
> > +       snprintf(buf, sizeof(buf), fmt ?: "", val);
> 
> Hi,
> 
> I couldn't reproduce this error and the code immediately above here
> does null check fmt:
> https://git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools-next.git/tree/tools/perf/util/stat-display.c?h=perf-tools-next#n437

Hi Ian,

I apologize for my carelessness. You're right, the code above doesn't need any change.
I'll send a new patch to fix it.

> Perhaps this is an older version of the perf tool? You can run "perf
> version" to determine this.

I found this bug with perf version 6.4.4, and I can reproduce the error with perf version 6.5.rc2.g76efcf004289.
The root cause is when I run `perf stat` like this:

  $ perf stat --metric-only -x, -e instructions -- sleep 1

It'll hit this line:
https://git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools-next.git/tree/tools/perf/util/stat-shadow.c#n318

The error can also be triggered by:
- Replace the `-x,` option with `-j`
- Replace the `-e instructions` option with `-e cycles`

Thanks,
Kaige

> [...]