Re: [PATCH 4/5] perf tools: Introduce perf_callchain_config()

[Namhyung Kim <namhyung@kernel.org>]

2014-09-22 (월), 16:24 +0200, Jiri Olsa:
> On Sun, Sep 21, 2014 at 01:18:05AM +0900, Namhyung Kim wrote:
> 
> SNIP
> 
> > +int perf_callchain_config(const char *var, const char *value)
> > +{
> > +	if (prefixcmp(var, "call-graph."))
> > +		return 0;
> > +	var += 11; /* strlen("call-graph.") == 11 */
> > +
> > +	if (!strcmp(var, "record-mode"))
> > +		return parse_callchain_record_opt(value);
> > +#ifdef HAVE_DWARF_UNWIND_SUPPORT
> > +	if (!strcmp(var, "dump-size")) {
> > +		unsigned long size = 0;
> > +		int ret;
> > +
> > +		ret = get_stack_size(value, &size);
> > +		callchain_param.dump_size = size;
> > +
> > +		return ret;
> > +	}
> > +#endif
> > +	if (!strcmp(var, "print-type"))
> > +		return parse_callchain_report_opt(value);
> > +	if (!strcmp(var, "order"))
> > +		return parse_callchain_report_opt(value);
> > +	if (!strcmp(var, "threshold"))
> > +		return parse_callchain_report_opt(value);
> > +	if (!strcmp(var, "print-limit"))
> > +		return parse_callchain_report_opt(value);
> > +	if (!strcmp(var, "sort-key"))
> > +		return parse_callchain_report_opt(value);
> 
> so this allows to spcify option like:
> 
> [call-graph]
>         sort-key = caller,function,1
> 
> which might be confusing for someone.. personaly I'm ok with that,
> I'm just raising it, because it's not apparent and not sure how
> strict we want to be in here ;-)

Right, I just reused parse_callchain_{record,report}_opt() functions to
simplify the code.

Hmm.. now I rethink that at least threshold and print-limit will be
treated differently as parse_callchain_report_opt() implies an order of
them.  Will change in the next version.

Thanks,
Namhyung