[PATCH 0/2] mm/damon/core: performance optimizations for kdamond hot path

[Josh Law <objecting@objecting.org>]

This patch series provides two performance optimizations for the DAMON
core, focusing on reducing unnecessary iterations and eliminating
expensive hardware integer divisions in the hot paths.

The first patch inverts the nested loops in kdamond_apply_schemes().
Previously, it iterated over all regions and then all schemes for each
region. By iterating over schemes on the outside, we can evaluate
scheme-level invariants (like activation status and quotas) once per
scheme and bypass the O(Regions) walk entirely when a scheme is inactive
or has fulfilled its quota.

The second patch eliminates a hardware integer division in
damon_max_nr_accesses(), which is called for every region during every
sample interval. It replaces the division with a read from a cached
field.

Together, these changes significantly reduce the CPU overhead of DAMON,
especially when monitoring large numbers of regions or using many
DAMOS schemes.

Josh Law (2):
  mm/damon/core: optimize kdamond_apply_schemes() by inverting scheme
    and region loops
  mm/damon/core: eliminate hot-path integer division in
    damon_max_nr_accesses()

 include/linux/damon.h |  3 +-
 mm/damon/core.c       | 72 ++++++++++++++++++-------------------------
 2 files changed, 31 insertions(+), 44 deletions(-)

-- 
2.34.1