Hi,
This series include a minor optimization of cpu partial slab fastpath,
which directly load freelist from cpu partial slab in the likely case.
It has small performance improvement in testing:
perf bench sched messaging -g 5 -t -l 100000
mm-stable slub-optimize
Total time 7.473 7.209
The other two patches are cleanups, which are included for convenience.
Thanks for review and comment!
To: Christoph Lameter <cl@xxxxxxxxx>
To: Pekka Enberg <penberg@xxxxxxxxxx>
To: David Rientjes <rientjes@xxxxxxxxxx>
To: Joonsoo Kim <iamjoonsoo.kim@xxxxxxx>
To: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
To: Vlastimil Babka <vbabka@xxxxxxx>
To: Roman Gushchin <roman.gushchin@xxxxxxxxx>
To: Hyeonggon Yoo <42.hyeyoo@xxxxxxxxx>
Cc: linux-mm@xxxxxxxxx
Cc: linux-kernel@xxxxxxxxxxxxxxx
Signed-off-by: Chengming Zhou <zhouchengming@xxxxxxxxxxxxx>
---
Chengming Zhou (3):
mm/slub: directly load freelist from cpu partial slab in the likely case
mm/slub: remove full list manipulation for non-debug slab
mm/slub: remove unused parameter in next_freelist_entry()
mm/slub.c | 31 +++++++++++++------------------
1 file changed, 13 insertions(+), 18 deletions(-)
---
base-commit: ab27740f76654ed58dd32ac0ba0031c18a6dea3b
change-id: 20240117-slab-misc-5a5f37a51257
Best regards,
--
Chengming Zhou <zhouchengming@xxxxxxxxxxxxx>
