On Tue, 17 Mar 2015 17:22:46 +0900 Roman Peniaev [off-list ref] wrote:

quoted

quoted

My second patch fixes this problem.
I occupy the block on allocation and avoid jumping to the search loop.

I'm not sure that this fixes above case.
'vm_map_ram (3) * 85' means 85 times vm_map_ram() calls.

First vm_map_ram(3) caller could get benefit from your second patch.
But, second caller and the other callers in each iteration could not
get benefit and should iterate whole list to find suitable free block,
because this free block is put to the tail of the list. Am I missing
something?

You are missing the fact that we occupy blocks in 2^n.
So in your example 4 page slots will be occupied (order is 2), not 3.

Could you please

- update the changelogs so they answer the questions which Joonsoo
  Kim and Gioh Kim asked

- write a little in-kernel benchmark to test the scenario which
  Joonsoo described and include the before and after timing results in
  the changelogs

- resend the patchset

Thanks.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>