Re: [patch] convert aio event reap to use atomic-op instead of spin_lock

[Benjamin LaHaise <bcrl@kvack.org>]

On Thu, Apr 12, 2007 at 12:50:39AM -0700, Ken Chen wrote:
> I ran through the autotest (with bug fix in the test code).  It passes
> the regression tests. I made the following change since last rev:

By removing the spinlock around ring insertion, you've made it possible 
for two events being inserted on different CPUs to end up creating 
inconsistent state, as there is nothing which guarantees that resulting 
event in the ring will be wholely one event or another.

Also, I don't like rounding up of the number of events to the nearest power 
of two, as it can round things up significantly more than just going to the 
end of the last actual page of memory allocated.

NAK.

		-ben

PS  base64 encoded patches are incredibly difficult to quote.
-- 
"Time is of no importance, Mr. President, only life is important."
Don't Email: <zyntrop@kvack.org>.