Re: [PATCH] tracing: Fix use-after-free and double-free on last_cmd

[Tze-nan Wu (吳澤南)]

On Sat, 2023-03-18 at 14:35 -0400, Steven Rostedt wrote:
> On Fri, 17 Mar 2023 13:30:44 +0800
> Cheng-Jui Wang <cheng-jui.wang@xxxxxxxxxxxx> wrote:
> 
> > From: "Tze-nan Wu" <Tze-nan.Wu@xxxxxxxxxxxx>
> 
> Hi!
> 
> Thanks for the report and the patch. Some nits below.
> 
> Also change the subject to:
> 
> 	tracing/synthetic: Fix races on freeing last_cmd
> 
> > ---
> >  kernel/trace/trace_events_synth.c | 23 +++++++++++++++++++----
> >  1 file changed, 19 insertions(+), 4 deletions(-)
> > 
> > diff --git a/kernel/trace/trace_events_synth.c
> > b/kernel/trace/trace_events_synth.c
> > index 46d0abb32d0f..ce438eccab2e 100644
> > --- a/kernel/trace/trace_events_synth.c
> > +++ b/kernel/trace/trace_events_synth.c
> > @@ -42,16 +42,25 @@ enum { ERRORS };
> >  #undef C
> >  #define C(a, b)		b
> >  
> > +static DEFINE_MUTEX(lastcmd_mutex);
> > +
> >  static const char *err_text[] = { ERRORS };
> >  
> >  static char *last_cmd;
> 
> Please keep the mutex and the variable it protects next to each
> other:
> 
> static DEFINE_MUTEX(lastcmd_mutex);
> static char *last_cmd;
> 
> >  
> >  static int errpos(const char *str)
> >  {
> > -	if (!str || !last_cmd)
> > -		return 0;
> > +	int ret = 0;
> > +
> > +	mutex_lock(&lastcmd_mutex);
> > +	if (!str || !last_cmd) {
> 
> Change this to just:
> 
> 	if (!str || !last_cmd)
> 		goto out;
> 
> > +		mutex_unlock(&lastcmd_mutex);
> > +		return ret;
> > +	}
> >  
> > -	return err_pos(last_cmd, str);
> > +	ret = err_pos(last_cmd, str);
> 
> Add:
> 
>  out:
> 
> > +	mutex_unlock(&lastcmd_mutex);
> > +	return ret;
> >  }
> >  
> >  static void last_cmd_set(const char *str)
> > @@ -59,18 +68,24 @@ static void last_cmd_set(const char *str)
> >  	if (!str)
> >  		return;
> >  
> > +	mutex_lock(&lastcmd_mutex);
> >  	kfree(last_cmd);
> >  
> 
> In this case, you can remove the space:
> 
> 	mutex_lock(&lastcmd_mutex);
> 	kfree(last_cmd);
> 	last_cmd = kstrdup(str, GFP_KERNEL);
> 	mutex_unlock(&lastcmd_mutex);
> 
> >  	last_cmd = kstrdup(str, GFP_KERNEL);
> > +	mutex_unlock(&lastcmd_mutex);
> >  }
> >  
> >  static void synth_err(u8 err_type, u16 err_pos)
> >  {
> > -	if (!last_cmd)
> > +	mutex_lock(&lastcmd_mutex);
> > +	if (!last_cmd) {
> 
> This should be:
> 
> 	if (!last_cmd)
> 		goto out;
> 
> > +		mutex_unlock(&lastcmd_mutex);
> >  		return;
> > +	}
> >  
> >  	tracing_log_err(NULL, "synthetic_events", last_cmd, err_text,
> >  			err_type, err_pos);
> 
>  out:
> 
> > +	mutex_unlock(&lastcmd_mutex);
> >  }
> >  
> >  static int create_synth_event(const char *raw_command);
> 
> Thanks,
> 
> -- Steve

Hi Steve,

Thanks for the comments,
the new patch with cleaner code is now ready.

Since the topic has been changed, I created a new thread for the new
patch:

https://lore.kernel.org/lkml/20230321110444.1587-1-Tze-nan.Wu@xxxxxxxxxxxx/

-- Tze-nan