Hello,
This series mainly optimizes cmp_and_merge_page() to have more efficient
separate code flow for ksm page and non-ksm anon page.
- ksm page: don't need to calculate the checksum obviously.
- anon page: don't need to search stable tree if changing fast and try
to merge with zero page before searching ksm page on stable tree.
Please see the patch-2 for details.
Patch-3 is cleanup also a little optimization for the chain()/chain_prune
interfaces, which made the stable_tree_search()/stable_tree_insert() over
complex.
In patch-4, fix behaviors in stable_tree_search() when handle migrating
stable_node: return the migrated ksm page if no shareable ksm page found
on the stable tree, so our rmap_item can be added directly.
I have done simple testing using "hackbench -g 1 -l 300000" (maybe I need
to use a better workload) on my machine, have seen a little CPU usage
decrease of ksmd and some improvements of cmp_and_merge_page() latency:
Before:
- ksm page
[128, 256) 21 | |
[256, 512) 12509 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@|
[512, 1K) 769 |@@@ |
[1K, 2K) 99 | |
[2K, 4K) 4 | |
[4K, 8K) 2 | |
[8K, 16K) 8 | |
- anon page
[512, 1K) 19 | |
[1K, 2K) 7160 |@@@@@@@@@@@ |
[2K, 4K) 33516 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@|
[4K, 8K) 33172 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ |
[8K, 16K) 11305 |@@@@@@@@@@@@@@@@@ |
[16K, 32K) 1303 |@@ |
[32K, 64K) 16 | |
[64K, 128K) 6 | |
[128K, 256K) 6 | |
[256K, 512K) 9 | |
[512K, 1M) 3 | |
[1M, 2M) 2 | |
[2M, 4M) 1 | |
After:
- ksm page
[128, 256) 9 | |
[256, 512) 915 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@|
[512, 1K) 41 |@@ |
[1K, 2K) 1 | |
[2K, 4K) 1 | |
- anon page
[512, 1K) 374 | |
[1K, 2K) 5367 |@@@@ |
[2K, 4K) 64362 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@|
[4K, 8K) 27721 |@@@@@@@@@@@@@@@@@@@@@@ |
[8K, 16K) 1047 | |
[16K, 32K) 63 | |
[32K, 64K) 7 | |
[64K, 128K) 6 | |
[128K, 256K) 5 | |
[256K, 512K) 3 | |
[512K, 1M) 1 | |
We can see the latency of cmp_and_merge_page() when handling non-ksm
anon pages has been improved.
Thanks for review and comments!
Signed-off-by: Chengming Zhou <chengming.zhou@xxxxxxxxx>
---
Chengming Zhou (4):
mm/ksm: refactor out try_to_merge_with_zero_page()
mm/ksm: don't waste time searching stable tree for fast changing page
mm/ksm: optimize the chain()/chain_prune() interfaces
mm/ksm: use ksm page itself if no another ksm page is found on stable tree
mm/ksm.c | 266 ++++++++++++++++++++-------------------------------------------
1 file changed, 84 insertions(+), 182 deletions(-)
---
base-commit: 2218eca02bc4203f68b8fb7e1116e5a2601506d1
change-id: 20240524-b4-ksm-scan-optimize-d2fd9401c357
Best regards,
--
Chengming Zhou <chengming.zhou@xxxxxxxxx>
