Re: [PATCH] tracing/filters: Improve subsystem filter

[Li Zefan <lizf@cn.fujitsu.com>]

>> -static void filter_free_subsystem_preds(struct event_subsystem *system)
>> +/*
>> + * flag == 0: remove all events' filter
>> + * flag == 1: clear filter->no_reset
>> + * flag == 2: remove all preds with no_reset == false
>> + */
> 
> Once an option overlap the boolean binary range, it's better
> to start thinking about an enum type, for better self-explaining code.
> 

Ok.

I was lazy to think of proper names for those enums..

>> +static void filter_free_subsystem_preds(struct event_subsystem *system,
>> +					int flag)
>>  {
>>  	struct ftrace_event_call *call;
>>  
>> @@ -428,6 +434,14 @@ static void filter_free_subsystem_preds(struct event_subsystem *system)
>>  		if (!call->define_fields)
>>  			continue;
>>  
>> +		if (flag == 1) {
>> +			call->filter->no_reset = false;
>> +			continue;
>> +		}
>> +
>> +		if (flag == 2 && call->filter->no_reset == true)
> 
> Or simply if (flag == 2 && call->filter->no_reset)
> 

Sure.

>> +			continue;
>> +
>>  		if (!strcmp(call->system, system->name)) {
>>  			filter_disable_preds(call);
>>  			remove_filter_string(call->filter);
>> @@ -529,7 +543,8 @@ static filter_pred_fn_t select_comparison_fn(int op, int field_size,
>>  
>>  static int filter_add_pred(struct filter_parse_state *ps,
>>  			   struct ftrace_event_call *call,
>> -			   struct filter_pred *pred)
>> +			   struct filter_pred *pred,
>> +			   bool apply)
> 
> 
> bool dry_run sounds perhaps more intuitive for what is happening there.

I guess you "try_run". ;)

> Because "apply" is surprising in a "add_thing" function, it's like
> a field to confirm that we know what what we are doing :)
> 
> (May be I start to become a PITA with my naming worries, especially
> since I'm usually not good at it in my patches :)
> 

I'll take "try_run". Naming is often hard for me.

> Otherwise, the rest looks good.
> 

Thanks!