[PATCH v6 1/4] tools/perf: support parsing parameterized events

[Jiri Olsa]

On Mon, Dec 22, 2014 at 11:30:45AM -0800, Sukadev Bhattiprolu wrote:
> Jiri Olsa [jolsa at redhat.com] wrote:
> | On Sun, Dec 21, 2014 at 11:49:24PM -0800, Sukadev Bhattiprolu wrote:
> | 
> | SNIP
> | 
> | > +	}
> | >  
> | >  	switch (format->value) {
> | >  	case PERF_PMU_FORMAT_VALUE_CONFIG:
> | > @@ -592,11 +629,16 @@ static int pmu_config_term(struct list_head *formats,
> | >  	}
> | >  
> | >  	/*
> | > -	 * XXX If we ever decide to go with string values for
> | > -	 * non-hardcoded terms, here's the place to translate
> | > -	 * them into value.
> | > +	 * Either directly use a numeric term, or try to translate string terms
> | > +	 * using event parameters.
> | >  	 */
> | > -	pmu_format_value(format->bits, term->val.num, vp, zero);
> | > +	if (term->type_val == PARSE_EVENTS__TERM_TYPE_NUM)
> | > +		val = term->val.num;
> | > +	else
> | > +		if (pmu_resolve_param_term(term, head_terms, &val))
> | > +			return -EINVAL;
> | > +
> | 
> | I'm ok with the change logic, but I'm missing here check for the 'term'
> | string value to be '?', so we force subst terms to have '?' as value..
> | I believe thats what we decided in the previous set discussion, right?
> 
> The =? is not a user input, so I did not think of validating that.
> 
> perf tool expects kernel/sysfs to show entries like 'core=?'. Are you
> saying that we should error out if kernel mistakenly displays 'core=$val'
> or 'core=?val' ? 

I think the we should at least try to have interface unambiguous
from the beginning

> If a required parameter is missing, we catch that in pmu_resolve_param_term().
> If a bogus parameter is specified we catch that above in pmu_config_term().

but the value of that param is unspecified, and if we later want to
add another type of string values we would be screwed.. as I described
in the previous reply for your other patch.

jirka