Re: [PATCH] tracing: save cmdline only when task does not exist in savecmd for optimization

[Yang Jihong]

Hi Steven,

On 2021/10/14 22:32, Steven Rostedt wrote:
> On Mon, 11 Oct 2021 19:50:18 +0800
> Yang Jihong <yangjihong1@xxxxxxxxxx> wrote:
>
> > diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
> > index 7896d30d90f7..a795610a3b37 100644
> > --- a/kernel/trace/trace.c
> > +++ b/kernel/trace/trace.c
> > @@ -2427,8 +2427,11 @@ static int trace_save_cmdline(struct task_struct *tsk)
> >   		savedcmd->cmdline_idx = idx;
> >   	}
> >   
> > -	savedcmd->map_cmdline_to_pid[idx] = tsk->pid;
> > -	set_cmdline(idx, tsk->comm);
> > +	/* save cmdline only when task does not exist in savecmd */
> > +	if (savedcmd->map_cmdline_to_pid[idx] != tsk->pid) {
> > +		savedcmd->map_cmdline_to_pid[idx] = tsk->pid;
> > +		set_cmdline(idx, tsk->comm);
> > +	}
> >   
>
> I now remember why I never did it this way. This breaks saving the command
> line when we do an exec.
>
> That is, just because mapped_pid == tsk->pid does not mean that the comm is
> the same.
If do an exec, the original process is replaced with a new binary. 
Therefore, the command changes but the tsk->pid does not change. 
Therefore, we need to savecmd here again?

Okay, I see. Thank you for the detailed explanation. :)

Thanks,
Jihong
> -- Steve
> .