On Mon, Dec 07, 2020 at 03:25:00PM -0300, Martin Pieuchot wrote:
> Simple diff below to add the rw_lock_held() as well as a read & write
> version.
> 
> This allows us to reduce the difference with NetBSD in UVM by adding the
> following checks:
> 
>       KASSERT(rw_write_held(amap->am_lock));
> 
> ok?
> 
> Index: sys/rwlock.h
> ===================================================================
> RCS file: /cvs/src/sys/sys/rwlock.h,v
> retrieving revision 1.26
> diff -u -p -r1.26 rwlock.h
> --- sys/rwlock.h      16 Jul 2019 01:40:49 -0000      1.26
> +++ sys/rwlock.h      7 Dec 2020 18:22:03 -0000
> @@ -168,6 +168,11 @@ int      rw_enter(struct rwlock *, int);
>  void rw_exit(struct rwlock *);
>  int  rw_status(struct rwlock *);
>  
> +#define      rw_read_held(rwl)       (rw_status(rwl) == RW_READ)
> +#define      rw_write_held(rwl)      (rw_status(rwl) == RW_WRITE)
> +#define      rw_lock_held(rwl)       (rw_write_held(rwl) || 
> rw_read_held(rwl))

I think rw_lock_held() should invoke rw_status() only once. This would
reduce the overhead. It could even be a proper C function.

> +
> +
>  void _rrw_init_flags(struct rrwlock *, const char *, int,
>           const struct lock_type *);
>  int  rrw_enter(struct rrwlock *, int);
> 

Reply via email to