Messages in this thread | | | Date | Tue, 20 Apr 2010 09:20:27 +0900 | Subject | Re: [PATCH 2/6] change alloc function in pcpu_alloc_pages | From | Minchan Kim <> |
| |
On Tue, Apr 20, 2010 at 2:45 AM, Christoph Lameter <cl@linux-foundation.org> wrote: > On Mon, 19 Apr 2010, Minchan Kim wrote: > >> Let's tidy my table. >> >> I made quick patch to show the concept with one example of pci-dma. >> (Sorry but I attach patch since web gmail's mangling.) >> >> On UMA, we can change alloc_pages with >> alloc_pages_exact_node(numa_node_id(),....) >> (Actually, the patch is already merged mmotm) > > UMA does not have the concept of nodes. Whatever node you specify is > irrelevant. Please remove the patch from mmotm.
I didn't change API name. The patch is just for optimization. http://lkml.org/lkml/2010/4/14/225 I think it's reasonable in UMA. Why do you want to remove it?
Do you dislike alloc_pages_exact_node naming? I added comment. http://lkml.org/lkml/2010/4/14/230 Do you think it isn't enough?
This patch results from misunderstanding of alloc_pages_exact_node. (http://marc.info/?l=linux-mm&m=127109064101184&w=2) At that time, I thought naming changing is worth. But many people don't like it. Okay. It was just trial and if everyone dislike, I don't have any strong cause. But this patch series don't relate to it. Again said, It's just for optimization patch.
Let's clarify other's opinion.
1. "I dislike alloc_pages_exact_node naming. Let's change it with more clear name." 2. "I hate alloc_pages_exact_node. It's trivial optimization. Let's remove it and replace it with alloc_pages_node." 3. "alloc_pages_exact_node naming is not bad. Let's add the comment to clear name" 4. "Let's cleanup alloc_pages_xxx in this change as well as 3. 5. "Please, don't touch. Remain whole of thing like as-is."
I think Chrsitop selects 5 or 1, Tejun selects 2, Mel selects 3, me want to 4 but is satisfied with 3. Right?
If we selects 5, In future, there are confusing between alloc_pages_node and alloc_pages_exact_node.So I don't want it.
If we select 2, We already have many place of alloc_pages_exact_node. And I add this patch series. So most of caller uses alloc_pages_exact_node now. Isn't it trivial?
So I want 3 at lest although you guys don't like 4. Please, suggest better idea to me. :)
-- Kind regards, Minchan Kim
| |