Re: [PATCH v2 32/40] tracing: Add cpu field for hist triggers

[Tom Zanussi <tom.zanussi@linux.intel.com>]

On Fri, 2017-09-08 at 15:08 -0400, Steven Rostedt wrote:
> On Tue,  5 Sep 2017 16:57:44 -0500
> Tom Zanussi <tom.zanussi@linux.intel.com> wrote:
> 
> > A common key to use in a histogram is the cpuid - add a new cpu
> > 'synthetic' field for that purpose.  This field is named cpu rather
> > than $cpu or $common_cpu because 'cpu' already exists as a special
> > filter field and it makes more sense to match that rather than add
> > another name for the same thing.
> > 
> > Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
> > ---
> >  Documentation/trace/events.txt   | 18 ++++++++++++++++++
> >  kernel/trace/trace_events_hist.c | 30 +++++++++++++++++++++++++++---
> >  2 files changed, 45 insertions(+), 3 deletions(-)
> > 
> > diff --git a/Documentation/trace/events.txt b/Documentation/trace/events.txt
> > index 2cc08d4..9717688 100644
> > --- a/Documentation/trace/events.txt
> > +++ b/Documentation/trace/events.txt
> > @@ -668,6 +668,24 @@ The following commands are supported:
> >    The examples below provide a more concrete illustration of the
> >    concepts and typical usage patterns discussed above.
> >  
> > +  'synthetic' event fields
> > +  ------------------------
> > +
> > +  There are a number of 'synthetic fields' available for use as keys
> > +  or values in a hist trigger.  These look like and behave as if they
> > +  were event fields, but aren't actually part of the event's field
> > +  definition or format file.  They are however available for any
> > +  event, and can be used anywhere an actual event field could be.
> > +  'Synthetic' field names are always prefixed with a '$' character to
> > +  indicate that they're not normal fields (with the exception of
> > +  'cpu', for compatibility with existing filter usage):
> > +
> > +    $common_timestamp      u64 - timestamp (from ring buffer) associated
> > +                                 with the event, in nanoseconds.  May be
> > +				 modified by .usecs to have timestamps
> > +				 interpreted as microseconds.
> 
> I guess the above should have been added with the synthetic field
> addition.
> 

Yeah, somehow this bit got shuffled to the wrong place refactoring the
patches. 

> > +    cpu                    int - the cpu on which the event occurred.
> 
> Then this (and the explanation of '$' for cpu above) should be added
> with this patch.
> 

Yep, will do.

> 
> > +
> >  
> >  6.2 'hist' trigger examples
> >  ---------------------------
> > diff --git a/kernel/trace/trace_events_hist.c b/kernel/trace/trace_events_hist.c
> > index 4f66f2e..0782766 100644
> > --- a/kernel/trace/trace_events_hist.c
> > +++ b/kernel/trace/trace_events_hist.c
> > @@ -226,6 +226,7 @@ enum hist_field_flags {
> >  	HIST_FIELD_FL_VAR_ONLY		= 8192,
> >  	HIST_FIELD_FL_EXPR		= 16384,
> >  	HIST_FIELD_FL_VAR_REF		= 32768,
> > +	HIST_FIELD_FL_CPU		= 65536,
> >  };
> >  
> >  struct var_defs {
> > @@ -1170,6 +1171,16 @@ static u64 hist_field_timestamp(struct hist_field *hist_field,
> >  	return ts;
> >  }
> >  
> > +static u64 hist_field_cpu(struct hist_field *hist_field,
> > +			  struct tracing_map_elt *elt,
> > +			  struct ring_buffer_event *rbe,
> > +			  void *event)
> > +{
> > +	int cpu = raw_smp_processor_id();
> 
> Hmm, I wonder if this should be just smp_processor_id(). Why have the
> raw_?
> 

You're right, smp_processor_id() should be fine.

> > +
> > +	return cpu;
> > +}
> > +
> >  static struct hist_field *check_var_ref(struct hist_field *hist_field,
> >  					struct hist_trigger_data *var_data,
> >  					unsigned int var_idx)
> > @@ -1518,6 +1529,8 @@ static const char *hist_field_name(struct hist_field *field,
> >  		field_name = hist_field_name(field->operands[0], ++level);
> >  	else if (field->flags & HIST_FIELD_FL_TIMESTAMP)
> >  		field_name = "$common_timestamp";
> > +	else if (field->flags & HIST_FIELD_FL_CPU)
> > +		field_name = "cpu";
> >  	else if (field->flags & HIST_FIELD_FL_EXPR ||
> >  		 field->flags & HIST_FIELD_FL_VAR_REF)
> >  		field_name = field->name;
> > @@ -1990,6 +2003,15 @@ static struct hist_field *create_hist_field(struct hist_trigger_data *hist_data,
> >  		goto out;
> >  	}
> >  
> > +	if (flags & HIST_FIELD_FL_CPU) {
> > +		hist_field->fn = hist_field_cpu;
> > +		hist_field->size = sizeof(int);
> > +		hist_field->type = kstrdup("int", GFP_KERNEL);
> > +		if (!hist_field->type)
> > +			goto free;
> > +		goto out;
> > +	}
> > +
> >  	if (WARN_ON_ONCE(!field))
> >  		goto out;
> >  
> > @@ -2182,7 +2204,9 @@ static struct hist_field *parse_var_ref(struct trace_array *tr,
> >  		hist_data->enable_timestamps = true;
> >  		if (*flags & HIST_FIELD_FL_TIMESTAMP_USECS)
> >  			hist_data->attrs->ts_in_usecs = true;
> > -	} else {
> > +	} else if (strcmp(field_name, "cpu") == 0)
> > +		*flags |= HIST_FIELD_FL_CPU;
> > +	else {
> >  		field = trace_find_event_field(file->event_call, field_name);
> >  		if (!field || !field->size) {
> >  			field = ERR_PTR(-EINVAL);
> 
> 
> > @@ -3185,7 +3209,6 @@ static int onmatch_create(struct hist_trigger_data *hist_data,
> >  				goto out;
> >  			}
> >  		}
> > -
> >  		if (param[0] == '$')
> >  			hist_field = onmatch_find_var(hist_data, data, system,
> >  						      event_name, param);
> > @@ -3200,7 +3223,6 @@ static int onmatch_create(struct hist_trigger_data *hist_data,
> >  			ret = -EINVAL;
> >  			goto out;
> >  		}
> > -
> 
> Why the modification of whitespace here?
> 

Again no good reason, another casualty of refactoring lots of
patches ;-)

Tom