Re: compile warning fix for advansys (trivial)

[Christoph Hellwig <hch@infradead.org>]

On Mon, Dec 29, 2003 at 12:26:39AM +0100, Guennadi Liakhovetski wrote:
> > It's not that easy.  You need to move the request_region as early as
> > the check_region is currently and make sure you properly release the
> > I/O region again on an initialization failure later on.
> 
> Hm, my assumption was, that if the old version, doing
> 
> if (check_region())
> 	fail();
> 
> if (!request_region())
> 	fail();
> 
> was correct, it would mean, that if check_region succeeds, but due to the
> race, the following request_region() fails, the error-processing and
> bailing-out should work correct. So, we can just safely assume, that the
> check_region() does succeed, and let the actual request_region decide. So,
> I think, the patch would only be wrong if either the original driver
> behaviour was wrong in the race-case,

it was.  That's why we're changing it after all :)