Messages in this thread | | | Date | Thu, 13 Dec 2007 18:54:32 -0500 | From | Kyle McMartin <> | Subject | Re: RFC: remove __read_mostly |
| |
On Thu, Dec 13, 2007 at 11:20:44PM +0100, Adrian Bunk wrote: > Is there anywhere in the kernel a case where __read_mostly brings a > measurable improvement or can it be removed? >
Yes, definitely[1]... the problem is, and this is also true of other annotations[2], that people go absolutely nuts adding these annotations without doing any profiling to see whether they actually improve things.
I'd bet, in the __read_mostly case at least, that there's no improvement in almost all cases.
cheers, Kyle
1. It's hugely relevant on big-SMP machines. 2. I mean __read_mostly, likely, unlikely, etc., not the useful sparse annotations.
| |