MIME-Version: 1.0
References: <20211117021223.2137117-1-kaleshsingh@google.com> <20211116224703.133f1750@rorschach.local.home>
In-Reply-To: <20211116224703.133f1750@rorschach.local.home>
From:   Kalesh Singh <kaleshsingh@google.com>
Date:   Tue, 16 Nov 2021 20:37:12 -0800
Message-ID: <CAC_TJveGzESpuzGWhyF+31=c4iLJ=TdNeivJyVf_Et-cyQUX3w@mail.gmail.com>
Subject: Re: [PATCH] tracing/histogram: Fix UAF in destroy_hist_field()
To:     Steven Rostedt <rostedt@goodmis.org>
Cc:     kernel-team@android.com, mhiramat@kernel.org, zanussi@kernel.org,
        kernel test robot <oliver.sang@intel.com>,
        Ingo Molnar <mingo@redhat.com>, linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Tue, Nov 16, 2021 at 7:47 PM Steven Rostedt <rostedt@goodmis.org> wrote:
>
> On Tue, 16 Nov 2021 18:12:22 -0800
> Kalesh Singh <kaleshsingh@google.com> wrote:
>
> > Calling destroy_hist_field() on an expression will recursively free
> > any operands associated with the expression. If during expression
> > parsing the operands of the expression are already set when an error
> > is encountered, there is no need to explicity free the operands. Doing
> > so will result in destroy_hist_field() being called twice for the
> > operands and lead to a use-after-free (UAF) error.
> >
> > Fix this by only calling destroy_hist_field() for the expression if the
> > operands are already set.
> >
> > Signed-off-by: Kalesh Singh <kaleshsingh@google.com>
> > Fixes: 8b5d46fd7a38 ("tracing/histogram: Optimize division by constants")
> > Reported-by: kernel test robot <oliver.sang@intel.com>
> > ---
> >  kernel/trace/trace_events_hist.c | 5 +++--
> >  1 file changed, 3 insertions(+), 2 deletions(-)
> >
> > diff --git a/kernel/trace/trace_events_hist.c b/kernel/trace/trace_events_hist.c
> > index 5ea2c9ec54a6..e3856eaf2ac3 100644
> > --- a/kernel/trace/trace_events_hist.c
> > +++ b/kernel/trace/trace_events_hist.c
> > @@ -2669,7 +2669,7 @@ static struct hist_field *parse_expr(struct hist_trigger_data *hist_data,
> >               if (!divisor) {
> >                       hist_err(file->tr, HIST_ERR_DIVISION_BY_ZERO, errpos(str));
> >                       ret = -EDOM;
> > -                     goto free;
> > +                     goto free_expr;
> >               }
> >
> >               /*
> > @@ -2709,7 +2709,7 @@ static struct hist_field *parse_expr(struct hist_trigger_data *hist_data,
> >               expr->type = kstrdup_const(operand1->type, GFP_KERNEL);
> >               if (!expr->type) {
> >                       ret = -ENOMEM;
> > -                     goto free;
> > +                     goto free_expr;
> >               }
> >
> >               expr->name = expr_str(expr, 0);
> > @@ -2719,6 +2719,7 @@ static struct hist_field *parse_expr(struct hist_trigger_data *hist_data,
> >  free:
> >       destroy_hist_field(operand1, 0);
> >       destroy_hist_field(operand2, 0);
> > +free_expr:
> >       destroy_hist_field(expr, 0);
> >
> >       return ERR_PTR(ret);
> >
> > base-commit: 8ab774587903771821b59471cc723bba6d893942
>
> Wouldn't this be a simpler and more robust fix?
>
> diff --git a/kernel/trace/trace_events_hist.c b/kernel/trace/trace_events_hist.c
> index 5ea2c9ec54a6..aab69b4ffe11 100644
> --- a/kernel/trace/trace_events_hist.c
> +++ b/kernel/trace/trace_events_hist.c
> @@ -2717,8 +2717,10 @@ static struct hist_field *parse_expr(struct hist_trigger_data *hist_data,
>
>         return expr;
>  free:
> -       destroy_hist_field(operand1, 0);
> -       destroy_hist_field(operand2, 0);
> +       if (!expr || expr->operand[0] != operand1)
> +               destroy_hist_field(operand1, 0);
> +       if (!expr || expr->operand[1] != operand2)
> +               destroy_hist_field(operand2, 0);
>         destroy_hist_field(expr, 0);

Hi Steve,

Agreed. What you suggested is simpler to work with. I'll respin a new version.

Thanks,
Kalesh

>
>         return ERR_PTR(ret);
>
>
> I'm worried about the complexity of having to know where to free what,
> and not just figuring it out at the end.
>
> -- Steve