Re: [PATCH] lock assertion macros for 2.5.30

[Jens Axboe]

On Thu, Aug 08 2002, Joshua MacDonald wrote:
> In order to implement MOST_NOT_HOLD_LOCK the spinlock must contain
> some record of who holds the lock, and since the SCSI-layer apparently

Correct

> does not have such a mechanism, it appears that something is broken in
> there.

I'll restate, the SCSI layer does _not_ use ASSERT_LOCK in the kernel!
If just one of the people that keep raising this point would actually
see what it does rather than assume, we would not keep seeing this
mentioned in this discussion!

-- 
Jens Axboe