On Sat, 2012-01-28 at 19:43 +0100, Jiri Olsa wrote:
> 
> +#define FILTER_TYPE_TRACE_FN           FILTER_TRACE_FN
> +#define FILTER_TYPE_TRACE_GRAPH_ENT    FILTER_OTHER
> +#define FILTER_TYPE_TRACE_GRAPH_RET    FILTER_OTHER
> +#define FILTER_TYPE_TRACE_CTX          FILTER_OTHER
> +#define FILTER_TYPE_TRACE_WAKE         FILTER_OTHER
> +#define FILTER_TYPE_TRACE_STACK                FILTER_OTHER
> +#define FILTER_TYPE_TRACE_USER_STACK   FILTER_OTHER
> +#define FILTER_TYPE_TRACE_BPRINT       FILTER_OTHER
> +#define FILTER_TYPE_TRACE_PRINT                FILTER_OTHER
> +#define FILTER_TYPE_TRACE_MMIO_RW      FILTER_OTHER
> +#define FILTER_TYPE_TRACE_MMIO_MAP     FILTER_OTHER
> +#define FILTER_TYPE_TRACE_BRANCH       FILTER_OTHER
> +#define FILTER_TYPE(arg)               FILTER_TYPE_##arg
> +
>  #undef FTRACE_ENTRY
>  #define FTRACE_ENTRY(name, struct_name, id, tstruct, print)		\

If all FTRACE_ENTRY needs a filter defined (as you did with the #defines
above), then we should just add a FILTER field to FTRACE_ENTRY(). The
defines are just ugly, and will be a pain if we ever add or remove a
type. If we remove a type, we'll probably forget to remove the define
for it.

#define FTRACE_ENTRY(name, struct_name, id, tstruct, print, filter)

	int filter_type = filter;

And then the trace entries can have the type of filter.

	F_STRUCT(
		...
	),

	FILTER_OTHER

That would be much cleaner.

-- Steve

>  int									\
> @@ -123,6 +137,7 @@ ftrace_define_fields_##name(struct ftrace_event_call *event_call)	\
>  {									\
>  	struct struct_name field;					\
>  	int ret;							\
> +	int filter_type = FILTER_TYPE(id);				\
>  									\
>  	tstruct;							\
>  									\