Re: [PATCH 05/14] perf tools: Introduce new 'ftrace' tool

[Jiri Olsa <jolsa@redhat.com>]

On Thu, Apr 25, 2013 at 03:06:37PM +0900, Namhyung Kim wrote:
> On Wed, 24 Apr 2013 16:09:18 +0200, Jiri Olsa wrote:
> > On Tue, Apr 23, 2013 at 05:31:03PM +0900, Namhyung Kim wrote:
> >> From: Namhyung Kim <namhyung.kim@lge.com>
> >> +out:
> >> +	put_tracing_file(file);
> >> +	return ret;
> >> +}
> >> +
> >> +static int reset_tracing_files(struct perf_ftrace *ftrace __maybe_unused)
> >> +{
> >> +	if (write_tracing_file("tracing_on", "0") < 0)
> >> +		return -1;
> >> +
> >> +	if (write_tracing_file("current_tracer", "nop") < 0)
> >> +		return -1;
> >> +
> >> +	if (write_tracing_file("set_ftrace_pid", " ") < 0)
> >> +		return -1;
> >
> > I think we want to reset set_ftrace_pid file as well
> > probably to -1..
> >
> > if I'm tracing command with this file having wrong CPU(s)
> > I dont get any data..
> 
> Not sure I understood what you mean exactly.

ugh, sry.. I wrote set_ftrace_pid when I actually meant tracing_cpumask ;-)

> 
> Do you mean it should be changed to
> 
> 	if (write_tracing_file("set_ftrace_pid", "-1") < 0)
> 
> ?  It seems it's the right thing to do for compatibility.


looks like it's enough only to open it with O_TRUNC,
so I guess whatever minimal goes write_tracing_file

---
kernel/trace/ftrace.c:

ftrace_pid_open(struct inode *inode, struct file *file)
{
        int ret = 0;

        if ((file->f_mode & FMODE_WRITE) &&
            (file->f_flags & O_TRUNC))
                ftrace_pid_reset();


jirka