Messages in this thread | | | Date | Sat, 13 Feb 2010 13:57:03 -0800 | From | "H. Peter Anvin" <> | Subject | Re: [PATCH 25/35] move round_up/down to kernel.h |
| |
On 02/13/2010 12:11 PM, Andrew Morton wrote: > >> Both of these names already exist -- yes, a rename is probably in order >> but it should be a separate patchset. > > More than a rename is certainly in order. > > And it's not clear who will be shovelling that muck, and when. We > haven't even worked out what needs to be done. >
OK, I was under the impression that all that was really called for was a rename -- which is pretty straightforward, after all; it can mostly be scripted.
I guess there is a bigger discussion here, and as such we should have it now. Since in the end, though, it is going to have to be a cross-kernel change, it is probably one of those things that should be done "just before -rc1".
Let's figure out what is needed. At a very minimum, we have the following operations:
-> align downward, arbitrary alignment -> align upward, arbitrary alignment - current roundup() -> align downward, power of 2 alignment - current round_down() -> align upward, power of 2 alignment - current round_up() -> truncate downward to a power of 2 - current rounddown_pow_of_two() -> truncate upward to a power of 2 - current roundup_pow_of_two()
Then we have some oddballs:
__ALIGN_MASK(x,m) == round_up(x, m-1) ALIGN(x,a) == round_up(x, m) PTR_ALIGN(p, a) == round_up with wrapper DIV_ROUND_UP(n,d) == roundup(n,d)/d DIV_ROUND_CLOSEST(n,d)
A structural renaming is certainly in order, but what more are you looking for, here?
-hpa
-- H. Peter Anvin, Intel Open Source Technology Center I work for Intel. I don't speak on their behalf.
| |