Re: [PATCH v3 3/9] perf stat: factor out event handling loop into a function

[Jiri Olsa]

On Wed, May 20, 2020 at 06:17:40PM +0300, Alexey Budankov wrote:

SNIP

> >> @@ -675,16 +708,9 @@ static int __run_perf_stat(int argc, const char **argv, int run_idx)
> >>  		perf_evlist__start_workload(evsel_list);
> >>  		enable_counters();
> >>  
> >> -		if (interval || timeout) {
> >> -			while (!waitpid(child_pid, &status, WNOHANG)) {
> >> -				nanosleep(&ts, NULL);
> >> -				if (timeout)
> >> -					break;
> >> -				process_interval();
> >> -				if (interval_count && !(--times))
> >> -					break;
> >> -			}
> >> -		}
> >> +		if (interval || timeout)
> >> +			handle_events(child_pid, &stat_config);
> >> +
> >>  		if (child_pid != -1) {
> >>  			if (timeout)
> >>  				kill(child_pid, SIGTERM);
> >> @@ -701,18 +727,7 @@ static int __run_perf_stat(int argc, const char **argv, int run_idx)
> >>  			psignal(WTERMSIG(status), argv[0]);
> >>  	} else {
> >>  		enable_counters();
> >> -		while (!done) {
> >> -			nanosleep(&ts, NULL);
> >> -			if (!is_target_alive(&target, evsel_list->core.threads))
> >> -				break;
> >> -			if (timeout)
> >> -				break;
> >> -			if (interval) {
> >> -				process_interval();
> >> -				if (interval_count && !(--times))
> >> -					break;
> >> -			}
> >> -		}
> >> +		handle_events(-1, &stat_config);
> > 
> > this makes me worried.. I'm not sure if it's good idea
> > to squash these 2 looops into one, because they are already
> > complex as they are.. and one of you following patches is
> > making it even more complex
> 
> Loops bodies are mostly identical. The only difference is in events
> they wait for and API used for that. Adding of more events will
> complicate further. The code is duplicated, thus needs refactoring.
> If the following patch complicates lets organize the patch it into
> several smaller functions.

yea, that might help

jirka

> 
> > 
> > wouldn't it be better if you just add single call into
> > each of them.. that would poll on your fd and process the
> > commands if needed?
> 
> That's of course possible, but doesn't manage existing complexity
> at the first place - __run_perf_stat().
> 
> Let's still have handle_events() as a general dispatcher and implement
> handlers for different events as separate functions?
> 
> ~Alexey
>