Re: [PATCH v2 2/2] trace-cmd: Reset CPU mask to its default value with "trace-cmd reset".

[Steven Rostedt <rostedt@goodmis.org>]

On Tue, 1 Oct 2019 18:37:54 +0300
Tzvetomir Stoyanov <tz.stoyanov@gmail.com> wrote:

> On Tue, Oct 1, 2019 at 6:32 PM Steven Rostedt <rostedt@goodmis.org> wrote:
> >
> > On Tue,  1 Oct 2019 17:57:40 +0300
> > "Tzvetomir Stoyanov (VMware)" <tz.stoyanov@gmail.com> wrote:
> >  
> > > "trace-cmd reset" command should put all ftrace config to its default
> > > state, but trace cpumask was not reseted. The patch sets cpumask to
> > > its default value - all CPUs are enabled for tracing.
> > >
> > > Signed-off-by: Tzvetomir Stoyanov (VMware) <tz.stoyanov@gmail.com>
> > > ---
> > >  tracecmd/trace-record.c | 35 +++++++++++++++++++++++++++++++++++
> > >  1 file changed, 35 insertions(+)
> > >
> > > diff --git a/tracecmd/trace-record.c b/tracecmd/trace-record.c
> > > index 69de82a..6f3520c 100644
> > > --- a/tracecmd/trace-record.c
> > > +++ b/tracecmd/trace-record.c
> > > @@ -4096,6 +4096,40 @@ static void reset_clock(void)
> > >               write_instance_file(instance, "trace_clock", "local", "clock");
> > >  }
> > >
> > > +static void reset_cpu_mask(void)
> > > +{
> > > +     struct buffer_instance *instance;
> > > +     int cpus = count_cpus();
> > > +     int fullwords;
> > > +     char *buf;
> > > +     int bits;
> > > +     int len;
> > > +
> > > +     fullwords = cpus / 32;
> > > +     bits = cpus % 32;
> > > +     len = (fullwords + 1) * 9;
> > > +
> > > +     buf = malloc(len + 1);
> > > +     if (!buf) {
> > > +             warning("Unable to allocate %d bytes for cpu mask", len + 1);
> > > +             return;
> > > +     }
> > > +     buf[0] = '\0';
> > > +     if (bits) {
> > > +             sprintf(buf, "%x", (1 << bits) - 1);
> > > +     } else if (fullwords) {
> > > +             strcat(buf, "ffffffff");
> > > +             fullwords--;
> > > +     }
> > > +     while (fullwords-- > 0)
> > > +             strcat(buf, ",ffffffff");  
> >
> > I like my way because it's a little more condensed ;-)
> >
> > But this works too and it's far from a fast path. As it's probably
> > more readable, I'll all this one.
> >
> > -- Steve
> >  
> Actually the only change here is to handle the case when bits is 0,
> not to put the leading ",".
> The ftrace complains with "Invalid argument" in that case.
> 

Mine handled that case too. The difference was the "cpus - 1" part of
my code:

	fullwords = (cpus - 1) / 32;  <--
	bits = (cpus - 1) % 32 + 1;   <--
	len = (fullwords + 1) * 9;

	buf = malloc(len + 1);
	buf[0] = '\0';

	if (bits)
		sprintf(buf, "%x", (unsigned int)((1ULL << bits) - 1));
	while (fullwords-- > 0)
		strcat(buf, ",ffffffff");


The output of yours and mine were identical. When cpus was a power of
32, it would still have one less of "fullwords", and bits would be 32.
Which is why I needed to add "1ULL" to make:

	(1 << 32) - 1 == 0xffffffff

the fullwords was only for non bits.

-- Steve