Re: spinlock assertion macros

[Arnd Bergmann]

On Wednesday 17 July 2002 04:22, Jesse Barnes wrote:
> files?  Anyway, I've got spinlock and rwlock versions of them below,
> maybe they're useful enough to go in as a start?  I only coded the
> ia64 version of rwlock_is_*_locked since it was easy--the i386
> versions were a little intimidating...
>
> I thought Oliver's suggestion for tracking the order of spinlock
> acquisition was good, hopefully someone will take a stab at it along
> with Dave's FUNCTION_SLEEPS() implementation.

I suppose you can simplify your interface when the code tracking the lock 
holder (i.e. the address of the lock call) is there: 

#define MUST_HOLD(lock)		BUG_ON(!(lock)->holder)

is independent of whether lock is a spinlock or an rw_lock, so you
don't need MUST_HOLD_READ anymore. I'd even go as far as dropping 
MUST_HOLD_WRITE as well, since it helps only in the corner case
where the lock is held but only for reading.

	Arnd <><