Re: [RESEND PATCH v10 08/25] dept: Apply sdt_might_sleep_{start,end}() to PG_{locked,writeback} wait

[Byungchul Park]

On Mon, Aug 21, 2023 at 05:09:44AM +0100, Matthew Wilcox wrote:
> On Mon, Aug 21, 2023 at 12:46:20PM +0900, Byungchul Park wrote:
> > @@ -1219,6 +1220,9 @@ static inline bool folio_trylock_flag(struct folio *folio, int bit_nr,
> >  /* How many times do we accept lock stealing from under a waiter? */
> >  int sysctl_page_lock_unfairness = 5;
> >  
> > +static struct dept_map __maybe_unused PG_locked_map = DEPT_MAP_INITIALIZER(PG_locked_map, NULL);
> > +static struct dept_map __maybe_unused PG_writeback_map = DEPT_MAP_INITIALIZER(PG_writeback_map, NULL);
> 
> Hmm, why are these "maybe unused"?  *digs*.  Ah.  Because
> sdt_might_sleep_start() becomes a no-op macro if DEPT is disabled.
> 
> OK, the right way to handle this is
> 
> #ifdef CONFIG_DEPT
> #define DEPT_MAP(name)	static struct dept_map name = \
> 		DEPT_MAP_INITIALIZER(name, NULL)
> #else
> #define DEPT_MAP(name)	/* */
> #endif
> 
> And now DEPT takes up no space if disabled.

Currently:

   #if !defined(CONFIG_DEPT)
   struct dept_map { }; 
   #endif

So I think it doesn't take space at all. Do you think I still need to
introduce e.g. DEPT_MAP()? Then I will. Thank you!

	Byungchul

> /* */; is a somewhat unusual thing to see, but since this must work at
> top level, we can't use "do { } while (0)" like we usually do.  Given
> where else this is likely to be used, i don't think it's going to be
> a problem ...