This is a follow-up to [1]: [PATCH v9 0/3] mm: process/cgroup ksm support which is not in mm-unstable yet (but soon? :) ). I'll be on vacation for ~2 weeks, so sending it out now as reply to [1]. (1) Make PR_SET_MEMORY_MERGE=0 unmerge pages like setting MADV_UNMERGEABLE does, (2) add a selftest for it and (3) factor out disabling of KSM from s390/gmap code. Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> Cc: Stefan Roesch <shr@xxxxxxxxxxxx> Cc: Rik van Riel <riel@xxxxxxxxxxx> Cc: Johannes Weiner <hannes@xxxxxxxxxxx> Cc: Michal Hocko <mhocko@xxxxxxxx> Cc: Christian Borntraeger <borntraeger@xxxxxxxxxxxxx> Cc: Janosch Frank <frankja@xxxxxxxxxxxxx> Cc: Claudio Imbrenda <imbrenda@xxxxxxxxxxxxx> Cc: Heiko Carstens <hca@xxxxxxxxxxxxx> Cc: Vasily Gorbik <gor@xxxxxxxxxxxxx> Cc: Sven Schnelle <svens@xxxxxxxxxxxxx> Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> Cc: Shuah Khan <shuah@xxxxxxxxxx> [1] https://lkml.kernel.org/r/20230418051342.1919757-1-shr@xxxxxxxxxxxx David Hildenbrand (3): mm/ksm: unmerge and clear VM_MERGEABLE when setting PR_SET_MEMORY_MERGE=0 selftests/ksm: ksm_functional_tests: add prctl unmerge test mm/ksm: move disabling KSM from s390/gmap code to KSM code arch/s390/mm/gmap.c | 20 +------ include/linux/ksm.h | 7 +++ kernel/sys.c | 7 +-- mm/ksm.c | 58 +++++++++++++++++++ .../selftests/mm/ksm_functional_tests.c | 46 +++++++++++++-- 5 files changed, 107 insertions(+), 31 deletions(-) -- 2.39.2
